Enum: FrontierModelEnum

Hierarchical enumeration of frontier AI models. The hierarchy has three tiers: developer (e.g. ANTHROPIC), model family (e.g. CLAUDE_OPUS), and specific model release identified by its canonical API model id (e.g. claude-opus-4-8). Intermediate developer and family nodes are uppercase categorical values; leaf nodes are the verbatim API identifiers.

URI: valuesets:FrontierModelEnum

View in BioPortal

Permissible Values

Value	Title	Meaning	Description	Api Model Id	Availability	Context Window	Developer	Full Model Id	Homepage	Input Usd Per Mtok	Knowledge Cutoff	License	Max Output Tokens	Node Type	Output Usd Per Mtok	Release Date	Specialization	Tier	Training Cutoff
ANTHROPIC			Models developed by Anthropic (the Claude family)						https://www.anthropic.com					developer
CLAUDE_OPUS			Claude Opus - Anthropic's most capable Opus-tier model family											family				most capable
claude-opus-4-8	Claude Opus 4.8		Claude Opus 4.8 - Anthropic's most capable Opus-tier model; highly autonomous, state-of-the-art on long-horizon agentic work, knowledge work, and memory. Adaptive thinking only.	claude-opus-4-8		1000000	Anthropic			5.0			128000	model	25.0
claude-opus-4-7	Claude Opus 4.7		Claude Opus 4.7 - previous-generation Opus; highly autonomous, strong on long-horizon agentic work, vision, and memory. Adaptive thinking only.	claude-opus-4-7		1000000	Anthropic						128000	model
claude-opus-4-6	Claude Opus 4.6		Claude Opus 4.6 - older Opus model supporting adaptive thinking and 128K max output tokens.	claude-opus-4-6		1000000	Anthropic						128000	model
claude-opus-4-5	Claude Opus 4.5		Claude Opus 4.5 - legacy Opus model (still active).	claude-opus-4-5			Anthropic	claude-opus-4-5-20251101						model
CLAUDE_SONNET			Claude Sonnet - Anthropic's balanced speed / intelligence family											family				balanced
claude-sonnet-4-6	Claude Sonnet 4.6		Claude Sonnet 4.6 - Anthropic's best combination of speed and intelligence. Supports adaptive thinking.	claude-sonnet-4-6		1000000	Anthropic			3.0			64000	model	15.0
claude-sonnet-4-5	Claude Sonnet 4.5		Claude Sonnet 4.5 - legacy Sonnet model (still active).	claude-sonnet-4-5			Anthropic	claude-sonnet-4-5-20250929						model
CLAUDE_HAIKU			Claude Haiku - Anthropic's fastest and most cost-effective family											family				fast
claude-haiku-4-5	Claude Haiku 4.5		Claude Haiku 4.5 - Anthropic's fastest and most cost-effective model for simple, latency-sensitive tasks.	claude-haiku-4-5		200000	Anthropic	claude-haiku-4-5-20251001		1.0			64000	model	5.0
CLAUDE_FABLE			Claude Fable - Anthropic's most capable widely released model family, for the most demanding reasoning and long-horizon agentic work.											family				frontier
claude-fable-5	Claude Fable 5		Claude Fable 5 - Anthropic's most capable widely released model. Thinking is always on; raw chain of thought is never returned. Requires 30-day data retention.	claude-fable-5		1000000	Anthropic			10.0			128000	model	50.0
CLAUDE_MYTHOS			Claude Mythos - same capabilities and API behavior as Claude Fable, available exclusively through Project Glasswing.		Project Glasswing only									family				frontier
claude-mythos-5	Claude Mythos 5		Claude Mythos 5 - same capabilities, pricing, limits, and API behavior as Claude Fable 5; available only through Project Glasswing.	claude-mythos-5		1000000	Anthropic						128000	model
OPENAI			Models developed by OpenAI (the GPT family)						https://openai.com					developer
GPT_5			GPT-5 - OpenAI's frontier model family for coding and professional work											family				frontier
gpt-5.5	GPT-5.5		GPT-5.5 - OpenAI's newest frontier model, "a new class of intelligence for coding and professional work".	gpt-5.5		1000000	OpenAI							model					2025-12-01
gpt-5.4	GPT-5.4		GPT-5.4 - a more affordable model for coding and professional work.	gpt-5.4		1000000	OpenAI							model					2025-08-31
gpt-5.4-mini	GPT-5.4 mini		GPT-5.4 mini - OpenAI's strongest mini model for coding, computer use, and subagents; lower latency and cost.	gpt-5.4-mini		400000	OpenAI							model					2025-08-31
gpt-5.2	GPT-5.2		GPT-5.2 - previous frontier model for complex professional work.	gpt-5.2			OpenAI							model
gpt-5.2-codex	GPT-5.2-Codex		GPT-5.2-Codex - coding-specialized variant for API-authenticated Codex workflows.	gpt-5.2-codex			OpenAI							model			coding
GOOGLE			Models developed by Google DeepMind (the Gemini family)						https://deepmind.google					developer
GEMINI_3			Gemini 3 - Google's latest generation multimodal model family											family				frontier
gemini-3.5-flash	Gemini 3.5 Flash		Gemini 3.5 Flash - Google's most intelligent model for agentic and coding tasks, delivering near-Pro intelligence at Flash-tier speed and cost.	gemini-3.5-flash		1000000	Google DeepMind							model		2026-05-19
gemini-3.1-pro-preview	Gemini 3.1 Pro		Gemini 3.1 Pro - reasoning-first model optimized for complex agentic workflows and coding, with adaptive thinking and integrated grounding.	gemini-3.1-pro-preview		1000000	Google DeepMind							model
gemini-3-flash-preview	Gemini 3 Flash		Gemini 3 Flash - frontier-class performance at reduced cost.	gemini-3-flash-preview			Google DeepMind							model
gemini-3.1-flash-lite	Gemini 3.1 Flash-Lite		Gemini 3.1 Flash-Lite - most cost-efficient Gemini model, optimized for low latency and high-volume traffic.	gemini-3.1-flash-lite			Google DeepMind							model
GEMINI_2_5			Gemini 2.5 - Google's previous-generation multimodal model family											family
gemini-2.5-pro	Gemini 2.5 Pro		Gemini 2.5 Pro - high-capability model for complex reasoning and coding with adaptive thinking and a 1M token context.	gemini-2.5-pro		1000000	Google DeepMind							model
gemini-2.5-flash	Gemini 2.5 Flash		Gemini 2.5 Flash - best price-performance for reasoning tasks.	gemini-2.5-flash			Google DeepMind							model
gemini-2.5-flash-lite	Gemini 2.5 Flash-Lite		Gemini 2.5 Flash-Lite - fastest and most budget-friendly option.	gemini-2.5-flash-lite			Google DeepMind							model
XAI			Models developed by xAI (the Grok family)						https://x.ai					developer
GROK_4			Grok 4 - xAI's reasoning-first flagship model family											family				frontier
grok-4.3	Grok 4.3		Grok 4.3 - xAI's reasoning-first flagship; the most intelligent and fastest model xAI has built.	grok-4.3		1000000	xAI							model		2026-04-30
grok-4.20-0309-reasoning	Grok 4.20 (reasoning)		Grok 4.20 (reasoning) - agentic tool-calling model noted for a very low hallucination rate.	grok-4.20-0309-reasoning		1000000	xAI							model
grok-4-0709	Grok 4		Grok 4 - earlier Grok 4 release.	grok-4-0709			xAI				2024-11			model
GROK_CODE			Grok Code - xAI's dedicated agentic coding model family											family			coding
grok-code-fast-1	Grok Code Fast 1		Grok Code Fast 1 - dedicated agentic coding model.	grok-code-fast-1			xAI							model			coding
META			Open-weight models developed by Meta (the Llama family)						https://www.llama.com			Llama Community License		developer
LLAMA_4			Llama 4 - Meta's open-weight mixture-of-experts model family											family
llama-4-maverick	Llama 4 Maverick		Llama 4 Maverick - general-purpose open-weight chat model; highest MMLU among Llama 4 variants.	llama-4-maverick			Meta							model
llama-4-scout	Llama 4 Scout		Llama 4 Scout - open-weight model supporting very long context (up to 10M tokens).	llama-4-scout		10000000	Meta							model
DEEPSEEK			Open-weight models developed by DeepSeek						https://www.deepseek.com					developer
DEEPSEEK_MODELS			DeepSeek - open-weight mixture-of-experts chat and reasoning models											family
deepseek-chat	DeepSeek-V3		DeepSeek-V3 (served as deepseek-chat) - open-weight MoE model activating ~37B of 671B parameters per token.	deepseek-chat			DeepSeek							model
deepseek-reasoner	DeepSeek-R1		DeepSeek-R1 (served as deepseek-reasoner) - open-weight reasoning model.	deepseek-reasoner			DeepSeek							model
MISTRAL			Models developed by Mistral AI						https://mistral.ai					developer
MISTRAL_LARGE			Mistral Large - Mistral AI's flagship open-weight model family											family
mistral-large-3	Mistral Large 3		Mistral Large 3 - state-of-the-art open-weight MoE model (41B active / 675B total parameters), 256K context, released under Apache 2.0.	mistral-large-3		256000	Mistral AI					Apache-2.0		model		2025-12
MOONSHOT			Models developed by Moonshot AI (the Kimi family)						https://www.moonshot.ai					developer
KIMI			Kimi - Moonshot AI's open-weight mixture-of-experts model family											family
kimi-k2.6	Kimi K2.6		Kimi K2.6 - frontier open-weight MoE coding model (32B active / 1T total parameters).	kimi-k2.6			Moonshot AI							model

Slots

Name	Description
frontier_model	A frontier AI model, its family, or its developer

Identifier and Mapping Information

Schema Source

• from schema: https://w3id.org/valuesets

LinkML Source

name: FrontierModelEnum
instantiates:
- valuesets_meta:ValueSetEnumDefinition
description: 'Hierarchical enumeration of frontier AI models. The hierarchy has three
  tiers: developer (e.g. ANTHROPIC), model family (e.g. CLAUDE_OPUS), and specific
  model release identified by its canonical API model id (e.g. claude-opus-4-8). Intermediate
  developer and family nodes are uppercase categorical values; leaf nodes are the
  verbatim API identifiers.'
title: Frontier AI Model
from_schema: https://w3id.org/valuesets
contributors:
- orcid:0000-0002-6601-2165
- https://github.com/anthropics/claude-code
status: DRAFT
rank: 1000
permissible_values:
  ANTHROPIC:
    text: ANTHROPIC
    description: Models developed by Anthropic (the Claude family)
    annotations:
      node_type:
        tag: node_type
        value: developer
      homepage:
        tag: homepage
        value: https://www.anthropic.com
    see_also:
    - https://docs.claude.com/en/docs/about-claude/models/overview
  CLAUDE_OPUS:
    text: CLAUDE_OPUS
    description: Claude Opus - Anthropic's most capable Opus-tier model family
    is_a: ANTHROPIC
    annotations:
      node_type:
        tag: node_type
        value: family
      tier:
        tag: tier
        value: most capable
  claude-opus-4-8:
    text: claude-opus-4-8
    description: Claude Opus 4.8 - Anthropic's most capable Opus-tier model; highly
      autonomous, state-of-the-art on long-horizon agentic work, knowledge work, and
      memory. Adaptive thinking only.
    is_a: CLAUDE_OPUS
    annotations:
      node_type:
        tag: node_type
        value: model
      api_model_id:
        tag: api_model_id
        value: claude-opus-4-8
      developer:
        tag: developer
        value: Anthropic
      context_window:
        tag: context_window
        value: 1000000
      max_output_tokens:
        tag: max_output_tokens
        value: 128000
      input_usd_per_mtok:
        tag: input_usd_per_mtok
        value: 5.0
      output_usd_per_mtok:
        tag: output_usd_per_mtok
        value: 25.0
    title: Claude Opus 4.8
    see_also:
    - https://docs.claude.com/en/docs/about-claude/models/overview
    aliases:
    - Opus 4.8
    - opus
  claude-opus-4-7:
    text: claude-opus-4-7
    description: Claude Opus 4.7 - previous-generation Opus; highly autonomous, strong
      on long-horizon agentic work, vision, and memory. Adaptive thinking only.
    is_a: CLAUDE_OPUS
    annotations:
      node_type:
        tag: node_type
        value: model
      api_model_id:
        tag: api_model_id
        value: claude-opus-4-7
      developer:
        tag: developer
        value: Anthropic
      context_window:
        tag: context_window
        value: 1000000
      max_output_tokens:
        tag: max_output_tokens
        value: 128000
    title: Claude Opus 4.7
    aliases:
    - Opus 4.7
  claude-opus-4-6:
    text: claude-opus-4-6
    description: Claude Opus 4.6 - older Opus model supporting adaptive thinking and
      128K max output tokens.
    is_a: CLAUDE_OPUS
    annotations:
      node_type:
        tag: node_type
        value: model
      api_model_id:
        tag: api_model_id
        value: claude-opus-4-6
      developer:
        tag: developer
        value: Anthropic
      context_window:
        tag: context_window
        value: 1000000
      max_output_tokens:
        tag: max_output_tokens
        value: 128000
    title: Claude Opus 4.6
    aliases:
    - Opus 4.6
  claude-opus-4-5:
    text: claude-opus-4-5
    description: Claude Opus 4.5 - legacy Opus model (still active).
    is_a: CLAUDE_OPUS
    annotations:
      node_type:
        tag: node_type
        value: model
      api_model_id:
        tag: api_model_id
        value: claude-opus-4-5
      full_model_id:
        tag: full_model_id
        value: claude-opus-4-5-20251101
      developer:
        tag: developer
        value: Anthropic
    title: Claude Opus 4.5
    aliases:
    - Opus 4.5
  CLAUDE_SONNET:
    text: CLAUDE_SONNET
    description: Claude Sonnet - Anthropic's balanced speed / intelligence family
    is_a: ANTHROPIC
    annotations:
      node_type:
        tag: node_type
        value: family
      tier:
        tag: tier
        value: balanced
  claude-sonnet-4-6:
    text: claude-sonnet-4-6
    description: Claude Sonnet 4.6 - Anthropic's best combination of speed and intelligence.
      Supports adaptive thinking.
    is_a: CLAUDE_SONNET
    annotations:
      node_type:
        tag: node_type
        value: model
      api_model_id:
        tag: api_model_id
        value: claude-sonnet-4-6
      developer:
        tag: developer
        value: Anthropic
      context_window:
        tag: context_window
        value: 1000000
      max_output_tokens:
        tag: max_output_tokens
        value: 64000
      input_usd_per_mtok:
        tag: input_usd_per_mtok
        value: 3.0
      output_usd_per_mtok:
        tag: output_usd_per_mtok
        value: 15.0
    title: Claude Sonnet 4.6
    aliases:
    - Sonnet 4.6
    - sonnet
  claude-sonnet-4-5:
    text: claude-sonnet-4-5
    description: Claude Sonnet 4.5 - legacy Sonnet model (still active).
    is_a: CLAUDE_SONNET
    annotations:
      node_type:
        tag: node_type
        value: model
      api_model_id:
        tag: api_model_id
        value: claude-sonnet-4-5
      full_model_id:
        tag: full_model_id
        value: claude-sonnet-4-5-20250929
      developer:
        tag: developer
        value: Anthropic
    title: Claude Sonnet 4.5
    aliases:
    - Sonnet 4.5
  CLAUDE_HAIKU:
    text: CLAUDE_HAIKU
    description: Claude Haiku - Anthropic's fastest and most cost-effective family
    is_a: ANTHROPIC
    annotations:
      node_type:
        tag: node_type
        value: family
      tier:
        tag: tier
        value: fast
  claude-haiku-4-5:
    text: claude-haiku-4-5
    description: Claude Haiku 4.5 - Anthropic's fastest and most cost-effective model
      for simple, latency-sensitive tasks.
    is_a: CLAUDE_HAIKU
    annotations:
      node_type:
        tag: node_type
        value: model
      api_model_id:
        tag: api_model_id
        value: claude-haiku-4-5
      full_model_id:
        tag: full_model_id
        value: claude-haiku-4-5-20251001
      developer:
        tag: developer
        value: Anthropic
      context_window:
        tag: context_window
        value: 200000
      max_output_tokens:
        tag: max_output_tokens
        value: 64000
      input_usd_per_mtok:
        tag: input_usd_per_mtok
        value: 1.0
      output_usd_per_mtok:
        tag: output_usd_per_mtok
        value: 5.0
    title: Claude Haiku 4.5
    aliases:
    - Haiku 4.5
    - haiku
  CLAUDE_FABLE:
    text: CLAUDE_FABLE
    description: Claude Fable - Anthropic's most capable widely released model family,
      for the most demanding reasoning and long-horizon agentic work.
    is_a: ANTHROPIC
    annotations:
      node_type:
        tag: node_type
        value: family
      tier:
        tag: tier
        value: frontier
  claude-fable-5:
    text: claude-fable-5
    description: Claude Fable 5 - Anthropic's most capable widely released model.
      Thinking is always on; raw chain of thought is never returned. Requires 30-day
      data retention.
    is_a: CLAUDE_FABLE
    annotations:
      node_type:
        tag: node_type
        value: model
      api_model_id:
        tag: api_model_id
        value: claude-fable-5
      developer:
        tag: developer
        value: Anthropic
      context_window:
        tag: context_window
        value: 1000000
      max_output_tokens:
        tag: max_output_tokens
        value: 128000
      input_usd_per_mtok:
        tag: input_usd_per_mtok
        value: 10.0
      output_usd_per_mtok:
        tag: output_usd_per_mtok
        value: 50.0
    title: Claude Fable 5
    aliases:
    - Fable 5
    - fable
  CLAUDE_MYTHOS:
    text: CLAUDE_MYTHOS
    description: Claude Mythos - same capabilities and API behavior as Claude Fable,
      available exclusively through Project Glasswing.
    is_a: ANTHROPIC
    annotations:
      node_type:
        tag: node_type
        value: family
      tier:
        tag: tier
        value: frontier
      availability:
        tag: availability
        value: Project Glasswing only
  claude-mythos-5:
    text: claude-mythos-5
    description: Claude Mythos 5 - same capabilities, pricing, limits, and API behavior
      as Claude Fable 5; available only through Project Glasswing.
    is_a: CLAUDE_MYTHOS
    annotations:
      node_type:
        tag: node_type
        value: model
      api_model_id:
        tag: api_model_id
        value: claude-mythos-5
      developer:
        tag: developer
        value: Anthropic
      context_window:
        tag: context_window
        value: 1000000
      max_output_tokens:
        tag: max_output_tokens
        value: 128000
    title: Claude Mythos 5
    aliases:
    - Mythos 5
  OPENAI:
    text: OPENAI
    description: Models developed by OpenAI (the GPT family)
    annotations:
      node_type:
        tag: node_type
        value: developer
      homepage:
        tag: homepage
        value: https://openai.com
    see_also:
    - https://developers.openai.com/api/docs/models
  GPT_5:
    text: GPT_5
    description: GPT-5 - OpenAI's frontier model family for coding and professional
      work
    is_a: OPENAI
    annotations:
      node_type:
        tag: node_type
        value: family
      tier:
        tag: tier
        value: frontier
  gpt-5.5:
    text: gpt-5.5
    description: GPT-5.5 - OpenAI's newest frontier model, "a new class of intelligence
      for coding and professional work".
    is_a: GPT_5
    annotations:
      node_type:
        tag: node_type
        value: model
      api_model_id:
        tag: api_model_id
        value: gpt-5.5
      developer:
        tag: developer
        value: OpenAI
      training_cutoff:
        tag: training_cutoff
        value: '2025-12-01'
      context_window:
        tag: context_window
        value: 1000000
    title: GPT-5.5
    see_also:
    - https://developers.openai.com/api/docs/models/gpt-5.5
    aliases:
    - GPT-5.5
  gpt-5.4:
    text: gpt-5.4
    description: GPT-5.4 - a more affordable model for coding and professional work.
    is_a: GPT_5
    annotations:
      node_type:
        tag: node_type
        value: model
      api_model_id:
        tag: api_model_id
        value: gpt-5.4
      developer:
        tag: developer
        value: OpenAI
      training_cutoff:
        tag: training_cutoff
        value: '2025-08-31'
      context_window:
        tag: context_window
        value: 1000000
    title: GPT-5.4
    aliases:
    - GPT-5.4
  gpt-5.4-mini:
    text: gpt-5.4-mini
    description: GPT-5.4 mini - OpenAI's strongest mini model for coding, computer
      use, and subagents; lower latency and cost.
    is_a: GPT_5
    annotations:
      node_type:
        tag: node_type
        value: model
      api_model_id:
        tag: api_model_id
        value: gpt-5.4-mini
      developer:
        tag: developer
        value: OpenAI
      training_cutoff:
        tag: training_cutoff
        value: '2025-08-31'
      context_window:
        tag: context_window
        value: 400000
    title: GPT-5.4 mini
    aliases:
    - GPT-5.4 mini
  gpt-5.2:
    text: gpt-5.2
    description: GPT-5.2 - previous frontier model for complex professional work.
    is_a: GPT_5
    annotations:
      node_type:
        tag: node_type
        value: model
      api_model_id:
        tag: api_model_id
        value: gpt-5.2
      developer:
        tag: developer
        value: OpenAI
    title: GPT-5.2
    aliases:
    - GPT-5.2
  gpt-5.2-codex:
    text: gpt-5.2-codex
    description: GPT-5.2-Codex - coding-specialized variant for API-authenticated
      Codex workflows.
    is_a: GPT_5
    annotations:
      node_type:
        tag: node_type
        value: model
      api_model_id:
        tag: api_model_id
        value: gpt-5.2-codex
      developer:
        tag: developer
        value: OpenAI
      specialization:
        tag: specialization
        value: coding
    title: GPT-5.2-Codex
    aliases:
    - GPT-5.2 Codex
  GOOGLE:
    text: GOOGLE
    description: Models developed by Google DeepMind (the Gemini family)
    annotations:
      node_type:
        tag: node_type
        value: developer
      homepage:
        tag: homepage
        value: https://deepmind.google
    see_also:
    - https://ai.google.dev/gemini-api/docs/models
  GEMINI_3:
    text: GEMINI_3
    description: Gemini 3 - Google's latest generation multimodal model family
    is_a: GOOGLE
    annotations:
      node_type:
        tag: node_type
        value: family
      tier:
        tag: tier
        value: frontier
  gemini-3.5-flash:
    text: gemini-3.5-flash
    description: Gemini 3.5 Flash - Google's most intelligent model for agentic and
      coding tasks, delivering near-Pro intelligence at Flash-tier speed and cost.
    is_a: GEMINI_3
    annotations:
      node_type:
        tag: node_type
        value: model
      api_model_id:
        tag: api_model_id
        value: gemini-3.5-flash
      developer:
        tag: developer
        value: Google DeepMind
      release_date:
        tag: release_date
        value: '2026-05-19'
      context_window:
        tag: context_window
        value: 1000000
    title: Gemini 3.5 Flash
    aliases:
    - Gemini 3.5 Flash
  gemini-3.1-pro-preview:
    text: gemini-3.1-pro-preview
    description: Gemini 3.1 Pro - reasoning-first model optimized for complex agentic
      workflows and coding, with adaptive thinking and integrated grounding.
    is_a: GEMINI_3
    annotations:
      node_type:
        tag: node_type
        value: model
      api_model_id:
        tag: api_model_id
        value: gemini-3.1-pro-preview
      developer:
        tag: developer
        value: Google DeepMind
      context_window:
        tag: context_window
        value: 1000000
    title: Gemini 3.1 Pro
    aliases:
    - Gemini 3.1 Pro
  gemini-3-flash-preview:
    text: gemini-3-flash-preview
    description: Gemini 3 Flash - frontier-class performance at reduced cost.
    is_a: GEMINI_3
    annotations:
      node_type:
        tag: node_type
        value: model
      api_model_id:
        tag: api_model_id
        value: gemini-3-flash-preview
      developer:
        tag: developer
        value: Google DeepMind
    title: Gemini 3 Flash
    aliases:
    - Gemini 3 Flash
  gemini-3.1-flash-lite:
    text: gemini-3.1-flash-lite
    description: Gemini 3.1 Flash-Lite - most cost-efficient Gemini model, optimized
      for low latency and high-volume traffic.
    is_a: GEMINI_3
    annotations:
      node_type:
        tag: node_type
        value: model
      api_model_id:
        tag: api_model_id
        value: gemini-3.1-flash-lite
      developer:
        tag: developer
        value: Google DeepMind
    title: Gemini 3.1 Flash-Lite
    aliases:
    - Gemini 3.1 Flash-Lite
  GEMINI_2_5:
    text: GEMINI_2_5
    description: Gemini 2.5 - Google's previous-generation multimodal model family
    is_a: GOOGLE
    annotations:
      node_type:
        tag: node_type
        value: family
  gemini-2.5-pro:
    text: gemini-2.5-pro
    description: Gemini 2.5 Pro - high-capability model for complex reasoning and
      coding with adaptive thinking and a 1M token context.
    is_a: GEMINI_2_5
    annotations:
      node_type:
        tag: node_type
        value: model
      api_model_id:
        tag: api_model_id
        value: gemini-2.5-pro
      developer:
        tag: developer
        value: Google DeepMind
      context_window:
        tag: context_window
        value: 1000000
    title: Gemini 2.5 Pro
    aliases:
    - Gemini 2.5 Pro
  gemini-2.5-flash:
    text: gemini-2.5-flash
    description: Gemini 2.5 Flash - best price-performance for reasoning tasks.
    is_a: GEMINI_2_5
    annotations:
      node_type:
        tag: node_type
        value: model
      api_model_id:
        tag: api_model_id
        value: gemini-2.5-flash
      developer:
        tag: developer
        value: Google DeepMind
    title: Gemini 2.5 Flash
    aliases:
    - Gemini 2.5 Flash
  gemini-2.5-flash-lite:
    text: gemini-2.5-flash-lite
    description: Gemini 2.5 Flash-Lite - fastest and most budget-friendly option.
    is_a: GEMINI_2_5
    annotations:
      node_type:
        tag: node_type
        value: model
      api_model_id:
        tag: api_model_id
        value: gemini-2.5-flash-lite
      developer:
        tag: developer
        value: Google DeepMind
    title: Gemini 2.5 Flash-Lite
    aliases:
    - Gemini 2.5 Flash-Lite
  XAI:
    text: XAI
    description: Models developed by xAI (the Grok family)
    annotations:
      node_type:
        tag: node_type
        value: developer
      homepage:
        tag: homepage
        value: https://x.ai
    see_also:
    - https://docs.x.ai/developers/models
  GROK_4:
    text: GROK_4
    description: Grok 4 - xAI's reasoning-first flagship model family
    is_a: XAI
    annotations:
      node_type:
        tag: node_type
        value: family
      tier:
        tag: tier
        value: frontier
  grok-4.3:
    text: grok-4.3
    description: Grok 4.3 - xAI's reasoning-first flagship; the most intelligent and
      fastest model xAI has built.
    is_a: GROK_4
    annotations:
      node_type:
        tag: node_type
        value: model
      api_model_id:
        tag: api_model_id
        value: grok-4.3
      developer:
        tag: developer
        value: xAI
      release_date:
        tag: release_date
        value: '2026-04-30'
      context_window:
        tag: context_window
        value: 1000000
    title: Grok 4.3
    aliases:
    - Grok 4.3
  grok-4.20-0309-reasoning:
    text: grok-4.20-0309-reasoning
    description: Grok 4.20 (reasoning) - agentic tool-calling model noted for a very
      low hallucination rate.
    is_a: GROK_4
    annotations:
      node_type:
        tag: node_type
        value: model
      api_model_id:
        tag: api_model_id
        value: grok-4.20-0309-reasoning
      developer:
        tag: developer
        value: xAI
      context_window:
        tag: context_window
        value: 1000000
    title: Grok 4.20 (reasoning)
    aliases:
    - Grok 4.20
  grok-4-0709:
    text: grok-4-0709
    description: Grok 4 - earlier Grok 4 release.
    is_a: GROK_4
    annotations:
      node_type:
        tag: node_type
        value: model
      api_model_id:
        tag: api_model_id
        value: grok-4-0709
      developer:
        tag: developer
        value: xAI
      knowledge_cutoff:
        tag: knowledge_cutoff
        value: 2024-11
    title: Grok 4
    aliases:
    - Grok 4
  GROK_CODE:
    text: GROK_CODE
    description: Grok Code - xAI's dedicated agentic coding model family
    is_a: XAI
    annotations:
      node_type:
        tag: node_type
        value: family
      specialization:
        tag: specialization
        value: coding
  grok-code-fast-1:
    text: grok-code-fast-1
    description: Grok Code Fast 1 - dedicated agentic coding model.
    is_a: GROK_CODE
    annotations:
      node_type:
        tag: node_type
        value: model
      api_model_id:
        tag: api_model_id
        value: grok-code-fast-1
      developer:
        tag: developer
        value: xAI
      specialization:
        tag: specialization
        value: coding
    title: Grok Code Fast 1
    aliases:
    - Grok Code Fast
  META:
    text: META
    description: Open-weight models developed by Meta (the Llama family)
    annotations:
      node_type:
        tag: node_type
        value: developer
      homepage:
        tag: homepage
        value: https://www.llama.com
      license:
        tag: license
        value: Llama Community License
    see_also:
    - https://www.llama.com/models/
  LLAMA_4:
    text: LLAMA_4
    description: Llama 4 - Meta's open-weight mixture-of-experts model family
    is_a: META
    annotations:
      node_type:
        tag: node_type
        value: family
  llama-4-maverick:
    text: llama-4-maverick
    description: Llama 4 Maverick - general-purpose open-weight chat model; highest
      MMLU among Llama 4 variants.
    is_a: LLAMA_4
    annotations:
      node_type:
        tag: node_type
        value: model
      api_model_id:
        tag: api_model_id
        value: llama-4-maverick
      developer:
        tag: developer
        value: Meta
    title: Llama 4 Maverick
    aliases:
    - Llama 4 Maverick
  llama-4-scout:
    text: llama-4-scout
    description: Llama 4 Scout - open-weight model supporting very long context (up
      to 10M tokens).
    is_a: LLAMA_4
    annotations:
      node_type:
        tag: node_type
        value: model
      api_model_id:
        tag: api_model_id
        value: llama-4-scout
      developer:
        tag: developer
        value: Meta
      context_window:
        tag: context_window
        value: 10000000
    title: Llama 4 Scout
    aliases:
    - Llama 4 Scout
  DEEPSEEK:
    text: DEEPSEEK
    description: Open-weight models developed by DeepSeek
    annotations:
      node_type:
        tag: node_type
        value: developer
      homepage:
        tag: homepage
        value: https://www.deepseek.com
    see_also:
    - https://api-docs.deepseek.com/
  DEEPSEEK_MODELS:
    text: DEEPSEEK_MODELS
    description: DeepSeek - open-weight mixture-of-experts chat and reasoning models
    is_a: DEEPSEEK
    annotations:
      node_type:
        tag: node_type
        value: family
  deepseek-chat:
    text: deepseek-chat
    description: DeepSeek-V3 (served as deepseek-chat) - open-weight MoE model activating
      ~37B of 671B parameters per token.
    is_a: DEEPSEEK_MODELS
    annotations:
      node_type:
        tag: node_type
        value: model
      api_model_id:
        tag: api_model_id
        value: deepseek-chat
      developer:
        tag: developer
        value: DeepSeek
    title: DeepSeek-V3
    aliases:
    - DeepSeek-V3
    - DeepSeek V3
  deepseek-reasoner:
    text: deepseek-reasoner
    description: DeepSeek-R1 (served as deepseek-reasoner) - open-weight reasoning
      model.
    is_a: DEEPSEEK_MODELS
    annotations:
      node_type:
        tag: node_type
        value: model
      api_model_id:
        tag: api_model_id
        value: deepseek-reasoner
      developer:
        tag: developer
        value: DeepSeek
    title: DeepSeek-R1
    aliases:
    - DeepSeek-R1
    - DeepSeek R1
  MISTRAL:
    text: MISTRAL
    description: Models developed by Mistral AI
    annotations:
      node_type:
        tag: node_type
        value: developer
      homepage:
        tag: homepage
        value: https://mistral.ai
    see_also:
    - https://docs.mistral.ai/getting-started/models/models_overview/
  MISTRAL_LARGE:
    text: MISTRAL_LARGE
    description: Mistral Large - Mistral AI's flagship open-weight model family
    is_a: MISTRAL
    annotations:
      node_type:
        tag: node_type
        value: family
  mistral-large-3:
    text: mistral-large-3
    description: Mistral Large 3 - state-of-the-art open-weight MoE model (41B active
      / 675B total parameters), 256K context, released under Apache 2.0.
    is_a: MISTRAL_LARGE
    annotations:
      node_type:
        tag: node_type
        value: model
      api_model_id:
        tag: api_model_id
        value: mistral-large-3
      developer:
        tag: developer
        value: Mistral AI
      release_date:
        tag: release_date
        value: 2025-12
      context_window:
        tag: context_window
        value: 256000
      license:
        tag: license
        value: Apache-2.0
    title: Mistral Large 3
    aliases:
    - Mistral Large 3
  MOONSHOT:
    text: MOONSHOT
    description: Models developed by Moonshot AI (the Kimi family)
    annotations:
      node_type:
        tag: node_type
        value: developer
      homepage:
        tag: homepage
        value: https://www.moonshot.ai
    see_also:
    - https://platform.moonshot.ai/docs
  KIMI:
    text: KIMI
    description: Kimi - Moonshot AI's open-weight mixture-of-experts model family
    is_a: MOONSHOT
    annotations:
      node_type:
        tag: node_type
        value: family
  kimi-k2.6:
    text: kimi-k2.6
    description: Kimi K2.6 - frontier open-weight MoE coding model (32B active / 1T
      total parameters).
    is_a: KIMI
    annotations:
      node_type:
        tag: node_type
        value: model
      api_model_id:
        tag: api_model_id
        value: kimi-k2.6
      developer:
        tag: developer
        value: Moonshot AI
    title: Kimi K2.6
    aliases:
    - Kimi K2.6