[
  {
    "id": "deepseek-v4-pro",
    "name": "DeepSeek V4 Pro (0813)",
    "lab": "DeepSeek",
    "release_date": "2026-08-13",
    "license": "open-weights",
    "price_in": 0.66,
    "price_out": 1.98,
    "price_source_url": "https://api-docs.deepseek.com/quick_start/pricing",
    "context_window": 1000000,
    "knowledge_cutoff": "",
    "lifecycle": "current",
    "superseded_by": null,
    "notes": "Off-peak pricing shown (01:00-04:00 and 06:00-10:00 UTC are peak hours, roughly 2x the off-peak rate). MIT-licensed weights on Hugging Face."
  },
  {
    "id": "claude-opus-4-8",
    "name": "Claude Opus 4.8",
    "lab": "Anthropic",
    "release_date": "2026-05-28",
    "license": "proprietary",
    "price_in": 5.0,
    "price_out": 25.0,
    "price_source_url": "https://www.anthropic.com/news/claude-opus-4-8",
    "context_window": 1000000,
    "knowledge_cutoff": "",
    "lifecycle": "superseded",
    "superseded_by": "claude-opus-5",
    "notes": "Superseded by Claude Opus 5 (2026-07-24, same $5/$25 pricing). Anthropic docs move it to 'Legacy models' with a migration guide, but it is formally still Active — tentative retirement not before 2027-05-28 — and serves as Claude Fable 5's automatic fallback for classifier-refused requests. Standard pricing shown; a $10/$50 'Fast Mode' tier is also offered."
  },
  {
    "id": "gpt-5-6-sol",
    "name": "GPT-5.6 Sol",
    "lab": "OpenAI",
    "release_date": "2026-07-09",
    "license": "proprietary",
    "price_in": 5.0,
    "price_out": 30.0,
    "price_source_url": "https://developers.openai.com/api/docs/pricing",
    "context_window": 1050000,
    "knowledge_cutoff": "2026-02-16",
    "lifecycle": "current",
    "superseded_by": null,
    "notes": "Standard tier shown; long-context tier above ~922K input tokens rises to $10/$45 per 1M tokens."
  },
  {
    "id": "gemini-3-1-pro-preview",
    "name": "Gemini 3.1 Pro Preview",
    "lab": "Google DeepMind",
    "release_date": "",
    "license": "proprietary",
    "price_in": 2.0,
    "price_out": 12.0,
    "price_source_url": "https://ai.google.dev/gemini-api/docs/pricing",
    "context_window": 1048576,
    "knowledge_cutoff": "",
    "lifecycle": "current",
    "superseded_by": null,
    "notes": "Pricing shown is for prompts <=200K context; rises to $4/$18 per 1M tokens beyond that. Still listed as Preview — official launch date and knowledge cutoff not found on the public model page."
  },
  {
    "id": "kimi-k3",
    "name": "Kimi K3",
    "lab": "Moonshot AI",
    "release_date": "2026-07-16",
    "license": "open-weights",
    "price_in": 3.0,
    "price_out": 15.0,
    "price_source_url": "https://platform.kimi.ai/docs/pricing/chat-k3",
    "context_window": 1048576,
    "knowledge_cutoff": "",
    "lifecycle": "current",
    "superseded_by": null,
    "notes": "Cache-hit input price is $0.30/1M, far below the $3.00 cache-miss price shown. Open weights under a bespoke license (not plain MIT/Apache) with commercial terms above certain revenue/MAU thresholds."
  },
  {
    "id": "qwen3-8-max",
    "name": "Qwen3.8-Max",
    "lab": "Alibaba",
    "release_date": "2026-08-03",
    "license": "proprietary",
    "price_in": 2.0,
    "price_out": 6.0,
    "price_source_url": "https://www.alibabacloud.com/help/en/model-studio/qwen3-8-max",
    "context_window": 1000000,
    "knowledge_cutoff": "",
    "lifecycle": "current",
    "superseded_by": null,
    "notes": "International/Singapore region pricing shown; China region pricing is lower ($1.65/$4.951 per 1M tokens). Knowledge cutoff not officially disclosed by Alibaba."
  },
  {
    "id": "glm-5-2",
    "name": "GLM-5.2",
    "lab": "Zhipu AI (Z.ai)",
    "release_date": "2026-06-16",
    "license": "open-weights",
    "price_in": 1.4,
    "price_out": 4.4,
    "price_source_url": "https://docs.z.ai/guides/overview/pricing",
    "context_window": 1000000,
    "knowledge_cutoff": "",
    "lifecycle": "current",
    "superseded_by": null,
    "notes": "MIT-licensed weights on Hugging Face and ModelScope. Cached-input price is $0.26/1M, far below the $1.40 shown. GLM-5.3 (2026-08-14) is positioned by Z.ai as the successor in this line, but is subscription-only with 'API coming soon' — GLM-5.2 remains the purchasable API model, so we keep it current until GLM-5.3 is actually available via API."
  },
  {
    "id": "claude-fable-5",
    "name": "Claude Fable 5",
    "lab": "Anthropic",
    "release_date": "2026-06-09",
    "license": "proprietary",
    "price_in": 10.0,
    "price_out": 50.0,
    "price_source_url": "https://platform.claude.com/docs/en/about-claude/models/overview",
    "context_window": 1000000,
    "knowledge_cutoff": "2026-01",
    "lifecycle": "current",
    "superseded_by": null,
    "notes": "First 'Mythos-class' model made generally available — a new tier above Opus, not a successor to Opus 4.8. Always-on adaptive thinking; safety classifiers can refuse requests (stop_reason 'refusal'), in which case Claude Opus 4.8 serves as automatic fallback — some published benchmark scores include those fallback answers. Suspended under US export controls 2026-06-12 to 06-30, restored 2026-07-01."
  },
  {
    "id": "claude-opus-5",
    "name": "Claude Opus 5",
    "lab": "Anthropic",
    "release_date": "2026-07-24",
    "license": "proprietary",
    "price_in": 5.0,
    "price_out": 25.0,
    "price_source_url": "https://platform.claude.com/docs/en/about-claude/models/overview",
    "context_window": 1000000,
    "knowledge_cutoff": "2026-05",
    "lifecycle": "current",
    "superseded_by": null,
    "notes": "Successor to Claude Opus 4.8 at the same $5/$25 price. New low/medium/high 'effort' toggle (defaults high). Anthropic positions it as 'close to the frontier intelligence of Claude Fable 5 at half the price'. Max output 128k (300k via batch beta)."
  },
  {
    "id": "glm-5-3",
    "name": "GLM-5.3",
    "lab": "Zhipu AI (Z.ai)",
    "release_date": "2026-08-14",
    "license": "proprietary at launch; open weights promised ~2 weeks post-launch pending safety review",
    "price_in": null,
    "price_out": null,
    "price_source_url": null,
    "context_window": 1000000,
    "knowledge_cutoff": null,
    "lifecycle": "current",
    "superseded_by": null,
    "notes": "No per-token API pricing yet — access is via GLM Coding Plan subscription ($18/$72/$160 per month) and ZCode, API 'coming soon'. Same 743B base model as GLM-5.2; all gains from scaled post-training (Z.ai claims +50% coding). Z.ai's own runs report DeepSWE v1.1 66.9 and Agents' Last Exam 28.5, but both used vendor harnesses that are not comparable to the independent leaderboards we track, so they are not listed as scores. Weights release delayed for safety evals after emergent exploit-chain capabilities."
  },
  {
    "id": "claude-sonnet-5",
    "name": "Claude Sonnet 5",
    "lab": "Anthropic",
    "release_date": "2026-06-30",
    "license": "proprietary",
    "price_in": 2.0,
    "price_out": 10.0,
    "price_source_url": "https://platform.claude.com/docs/en/about-claude/pricing",
    "context_window": 1000000,
    "knowledge_cutoff": "2026-01",
    "lifecycle": "current",
    "superseded_by": null,
    "notes": "Mid-tier of the Claude 5 family; Anthropic's docs place its predecessor Sonnet 4.6 in 'Legacy models'. Launch's introductory $2/$10 pricing was later made the standard price — the scheduled increase to $3/$15 will not occur, per the official pricing docs. Adaptive thinking with an effort control that defaults to high. Max output 128k."
  }
]