Last updated October 18, 2026

Models

Intelligence, capability, cost, and open versus closed.

Summary

This page tracks current progress on major AI models and how the field got here since 2022: independent leaderboard scores, vendor-published scores, list prices, and whether weights are open. The three sections below cover those themes with charts and commentary.

  • Intelligence & Capability. Frontier launch scores keep climbing on coding and capability evals, even as those tests get harder.
  • Capability vs cost. Open-weight models, especially from China, keep undercutting closed US list prices at useful capability.
  • Open vs closed. China leads open-weight volume among tracked labs. Published parameter counts for open-weight models now land in the same ballpark as the closed models that disclose size. That size overlap is one factor that helps explain why open models look closer on capability.

Formatted citation and BibTeX are in the Copy menu at the top of this page. The compiled table and original charts may be reused with attribution under a Creative Commons Attribution 4.0 license.

Jump to Section Contents

Intelligence & Capability

Four evals commonly cited for distinct capabilities: coding (FrontierCode), agent terminal work (Terminal-Bench), hard general knowledge (Humanity's Last Exam), and real work deliverables across 44 occupations, rated by Elo (GDPval-AA). Coding plots FrontierCode 1.1 Main. Moves on the chart are usually models improving or evals getting tougher.

Each lab's top leaderboard score over time

Legend

Source: FrontierCode

Figure 1. Each lab's top FrontierCode 1.1 Main score over time, as of Oct 16, 2026. Source: vendor announcements.

Table 1. FrontierCode launch scores, newest first.

Milestones

Dated launches that made a new kind of work practical. The same number on the axis and in the list is the same launch. Charts restart each year, except 2022 and 2023, which share a sequence because the industry was still early and gaining traction.

When those shifts shipped

Figure 2. Milestones on a linear time axis, with names in the list below. Source: vendor launch announcements.

2026

  1. Source control rebuilt for agents

    Anthropic, Scion. Agents began merging code faster than git setups built for people could handle. Cloudflare (April), GitLab (June) and GitHub (October) announced rebuilt source control for agents. Anthropic went from about 2,500 merged changes a day in May to over 30,000 in September and replaced git with its own system, Scion.

  2. Decision models spread after Jev

    TypeSafe AI, Jev. TypeSafe AI released Jev, a model that returns probabilities for typed yes/no, choice and score questions. Within a month Cloudflare , OpenAI , AWS 's Strands project and Ollama shipped their own decision models or Jev-compatible APIs.

  3. Computer use and coding become a launch claim on GPT-6 Astra

    OpenAI, GPT-6 Astra. OpenAI 's GPT-6 generation, launched first to Daybreak partners then paid ChatGPT; vendor-claimed step change on computer use, coding, and saturated ARC-AGI-3.

  4. Productized multi-bot teammates with a shared computer

    xAI, Grok Bot. Grok Bot beta. Always-on AI teammates that share a cloud computer, sign into the user's apps, message each other in threads or group chats, and hand work back only when approval is needed.

  5. Multi-agent execution becomes a launch feature

    OpenAI, GPT-5.6. First release split into three durable capability tiers, Sol (flagship), Terra, and Luna, each advancing on its own cadence; Sol Ultra multi-agent mode tops Terminal-Bench 2.1 at 91.9%.

  6. Days-long autonomous tasks become a shipping launch claim

    Anthropic, Claude Fable 5. Briefly suspended under USA export controls days after launch.

2025

  1. Long-running agent work becomes a launch feature

    Anthropic, Claude Sonnet 4.5. Billed as the best coding model on release, with 30-plus-hour autonomous agent runs.

  2. Instant and extended reasoning merge in one model

    Anthropic, Claude 3.7 Sonnet. First hybrid reasoning model, blending instant replies and extended thinking in one model; shipped with Claude Code.

  3. Open-weight reasoning reaches frontier-level public attention

    DeepSeek, DeepSeek-R1. MIT-licensed o1-class reasoning model whose release triggered a global market shock and a wave of RL replication.

2024

  1. Test-time reasoning becomes a paid product tier

    OpenAI, o1. First mainstream reasoning model, introducing chain-of-thought test-time compute as a product.

  2. Native multimodal interaction becomes a shipping product

    OpenAI, GPT-4o. Natively multimodal omni model; the launch demoed real-time voice, but the API shipped with text and image in, text out.

  3. Million-token context becomes a shipping model feature

    Google, Gemini 1.5 Pro. Made million-token context real, an order of magnitude beyond anything shipping at the time.

2022 / 2023

  1. Tool use becomes a shipping API feature

    OpenAI, Function calling. Developers could describe functions to GPT-4 and GPT-3.5-turbo. The model returned a JSON call with arguments.

  2. Consumer chat reaches the mainstream

    OpenAI, GPT-3.5 / ChatGPT. The ChatGPT launch that took LLMs mainstream and kicked off the current AI race.

Capability vs cost

Capability alone does not drive AI adoption, because the work also has to be cheap enough to run at scale. Plotting a launch's score against its list price is what people call a Pareto frontier: the best tradeoffs between capability and cost you can get at once.

Visualizing the Pareto frontier

Each line is the Pareto frontier for its snapshot, and only frontier models are plotted. Every other model falls in the shaded area, where it costs more for the same score or scores lower at the same price. The CSV lists every model.

FrontierCode score vs list input price

FrontierCode (%)

Other models sit in the shaded area: less capable at that price, or more expensive for that score.

Frontier models, Current
ModelCompanyInput list $/1MFrontierCode
Claude Haiku 5.5Anthropic0.146.4%
Claude Sonnet 5.5Anthropic252.1%
Claude Opus 5.5Anthropic454.6%

Legend

Figure 3. FrontierCode 1.1 Main score against list price ($/1M input), frontier models released by Oct 16, 2026. Source: Cognition FrontierCode, vendor prices.

Cheapest and highest-scoring launches on the line, by as-of date.
Oct 16, 2026Current$0.10 · 46.4%Claude Haiku 5.5$4 · 54.6%Claude Opus 5.5Live 1.1 Main board.
Sep 30, 2026Q3 2026$0.10 · 42.4%GPT-6 Luna$4 · 54.6%Claude Opus 5.5Quarter end 30 Sep 2026.
Jun 30, 2026Q2 2026$0.75 · 27%GPT-5.4-mini$10 · 53.5%Claude Fable 5FrontierCode 1.1 Main, quarter end Jun 30, 2026.

Table 2. Cheapest and highest-scoring launches on the frontier line, by as-of date; percentages are not comparable across FrontierCode boards.

Open vs closed major models

Tracked labs are still shipping fast in 2026. Open-weight launches keep landing, and China accounts for most of the open count in 2025-2026.

United States
13 open 40 closed
tracked launches, 2025-2026
China
35 open 6 closed
tracked launches, 2025-2026
France
3 open 2 closed
tracked launches, 2025-2026

Tracked releases by launch year

Launches

Legend

lighter = open, darker = closed

Figure 4. Open and closed launches by year, with the yearly total above each column; open includes later weight releases. Source: vendor launch announcements.

Table 3. Main model versions by region and year; open includes models with restricted licenses.

Open launches in 2026

Open-weight launches, including models with restricted licenses.

Open-weight scale

Most closed models, including Claude, Gemini, and ChatGPT, do not publish a parameter count. The points here are lab-published totals, plus xAI sizes Elon has stated. Open-weight counts have climbed from tens of billions into the trillions, and now sit in the same range as the Grok figures.

Parameter counts

Parameters

Legend

  • Open
  • Closed
  • Stated / planned
  • Trend

Figure 5. Parameter count over time, with landmark models labeled. Source: vendor announcements, model cards and Elon Musk.

Table 4. Parameter counts over time.

Launch table

Every tracked launch, newest first. The table opens on Frontier labs; switch to All companies for the rest.

Launch snapshots

Launch-state snapshots of major AI models
Launch benchmarks
2026-10-15 Muse Spark Open Meta open 1M n/a Terminal-Bench 2.1 76.4%*; GPQA Diamond 89.0%*; Humanity's Last Exam 43.9%
2026-10-15 Nemotron 3.5 Bolt 74B-A9B NVIDIA open 256k n/a Terminal-Bench 2.1 50.98%*; GPQA Diamond 86.43%*
2026-10-15 Nemotron 3.5 Super VL NVIDIA open n/a n/a n/a
2026-10-12 Muse Wonder Meta closed 1M $4 / $20 n/a
2026-10-12 Muse Wonder Full Meta closed 1M n/a Humanity's Last Exam 53.6%
2026-10-07 Claude Haiku 5.5 Anthropic closed 1M $0.10 / $0.50 FrontierCode 1.1 Main 46.4%; Humanity's Last Exam 44.4%
2026-10-06 Mistral Large 4 Preview Mistral AI closed n/a n/a FrontierCode 1.1 Main 39.4%
2026-09-30 Gemini 4 Argon Google closed n/a $2 / $10 FrontierCode 1.1 Main 47.5%; Terminal-Bench 4.0 57.4%*; Humanity's Last Exam 57.1%
2026-09-29 GPT-6.1 Sol OpenAI closed 1M $2 / $10 FrontierCode 1.1 Main 50.2%; Terminal-Bench 4.0 58.18%
2026-09-28 Claude Sonnet 5.5 Anthropic closed 1M $2 / $10 FrontierCode 1.1 Main 52.1%; Terminal-Bench 4.0 61.82%; Humanity's Last Exam 55.0%
2026-09-22 Claude Opus 5.5 Anthropic closed 1M $4 / $20 FrontierCode 1.1 Main 54.6%; Terminal-Bench 4.0 64.85%; Humanity's Last Exam 61.4%
2026-09-22 GPT-6 Luna OpenAI closed 1M $0.10 / $0.50 FrontierCode 1.1 Main 42.4%; Terminal-Bench 4.0 16.36%
2026-09-22 GPT-6 Sol OpenAI closed 1M $2 / $10 FrontierCode 1.1 Main 49.3%; Terminal-Bench 4.0 49.39%
2026-09-21 Grok 4.7 xAI closed 500k $2 / $6 FrontierCode 1.1 Main 47.6%; Terminal-Bench 4.0 37.58%
2026-09-21 MiMo-V2.6-Flash Xiaomi open 1M $0.14 / $0.28 n/a
2026-09-21 MiMo-V2.6-Pro Xiaomi open 1M $0.44 / $0.87 Artificial Analysis Intelligence Index 46 (vendor-claimed on Xiaomi page)
2026-09-10 DeepSeek-V4.1-Flash DeepSeek open n/a n/a FrontierCode 1.1 Main 43.9%
2026-09-03 GPT-6 Astra OpenAI closed 1M $10 / $50 FrontierCode 1.1 Main 53.3%; Terminal-Bench 4.0 58.18%; GPQA Diamond 96.0%; Humanity's Last Exam 54.7%
2026-09-02 Gemini 3.8 Flash Google closed 1M n/a FrontierCode 1.1 Main 41.2%; Terminal-Bench 4.0 19.09%; Humanity's Last Exam 47.8%
2026-09-01 Claude Fable 5.1 Anthropic closed 1M $10 / $50 FrontierCode 1.1 Main 50.9%; Terminal-Bench 4.0 57.88%; GPQA Diamond 93.7%; Humanity's Last Exam 59.1%
2026-08-27 Qwen3.8-Flash-Next Alibaba open n/a n/a n/a
2026-08-26 GLM-5.3-Flash Z.ai (Zhipu) open n/a n/a FrontierCode 1.1 Main 31.8%; Terminal-Bench 4.0 35.76%
2026-08-14 GLM-5.3 Z.ai (Zhipu) open 200k n/a FrontierCode 1.1 Main 40.1%; Terminal-Bench 4.0 41.82%
2026-08-13 DeepSeek V4 Pro 0813 DeepSeek open n/a n/a FrontierCode 1.1 Main 28.5%
2026-08-13 Gemini 3.7 Flash Google closed 1M n/a FrontierCode 1.1 Main 43.6%; Terminal-Bench 4.0 11.21%
2026-08-12 Grok 4.6 xAI closed 500k $2 / $6 FrontierCode 1.1 Main 48%; Terminal-Bench 4.0 20.3%
2026-08-11 Nemotron 3.5 Lightning NVIDIA open n/a n/a n/a
2026-08-02 Qwen3.8-Max Alibaba open n/a n/a Terminal-Bench 4.0 26.97%
2026-07-31 DeepSeek V4 Flash 0731 DeepSeek open n/a n/a FrontierCode 1.1 Main 18.8%
2026-07-24 Claude Opus 5 Anthropic closed 1M $5 / $25 FrontierCode 1.1 Main 53.4%; Humanity’s Last Exam 56.3%*; Terminal-Bench 4.0 53.94%
2026-07-21 Gemini 3.6 Flash Google closed 1M $1.50 / $7.50 FrontierCode 1.1 Main 34.4%
2026-07-16 Grok 4.5 xAI closed 500k $2 / $6 FrontierCode 1.1 Main 42.4%; Terminal-Bench 2.1 83.3%; Terminal-Bench 4.0 12.42%
2026-07-16 Kimi K3 Moonshot AI open n/a n/a FrontierCode 1.1 Main 44.2%
2026-07-15 Inkling Thinking Machines Lab open 1M n/a FrontierCode 1.1 Main 14%
2026-07-09 GPT-5.6 OpenAI closed 1M $4 / $20 GPQA Diamond 94.6%; Terminal-Bench 2.1 88.8%
2026-07-09 GPT-5.6 Luna OpenAI closed n/a n/a FrontierCode 1.1 Main 39.8%; Terminal-Bench 4.0 17.27%
2026-07-09 GPT-5.6 Sol OpenAI closed n/a n/a FrontierCode 1.1 Main 47.5%; Terminal-Bench 4.0 37.27%
2026-07-09 GPT-5.6 Terra OpenAI closed n/a n/a FrontierCode 1.1 Main 41.3%; Terminal-Bench 4.0 21.52%
2026-07-02 Hy3 Tencent open 256k n/a n/a
2026-06-30 Claude Sonnet 5 Anthropic closed 1M $2 / $10 FrontierCode 1.1 Main 42.7%; Humanity’s Last Exam 34.6% (no tools)*; Terminal-Bench 4.0 12.42%
2026-06-16 GLM-5.2 Z.ai (Zhipu) open 200k n/a FrontierCode 1.1 Main 24.5%
2026-06-09 Claude Fable 5 Anthropic closed 1M $10 / $50 FrontierCode 1.1 Main 53.5%; Terminal-Bench 2.1 88.0%; Terminal-Bench 4.0 44.55%; Humanity's Last Exam 55.5%
2026-06-04 Nemotron 3 Ultra NVIDIA open 1M n/a FrontierCode 1.1 Main 13.6%
2026-06-02 MiniMax-M3 MiniMax open 1M n/a FrontierCode 1.1 Main 14.7%
2026-05-28 Claude Opus 4.8 Anthropic closed 1M $5 / $25 FrontierCode 1.1 Main 46.5%; Terminal-Bench 4.0 23.64%
2026-05-22 Mistral Medium 3.5 Mistral AI open n/a n/a FrontierCode 1.1 Main 8%
2026-05-20 Kimi K2.7 Moonshot AI open n/a n/a FrontierCode 1.1 Main 30.1%
2026-05-19 Gemini 3.5 Flash Google closed 1M $1.50 / $9 n/a
2026-05-18 Qwen 3.7 Plus Alibaba open n/a n/a FrontierCode 1.1 Main 10.2%
2026-05-18 Qwen3.7-Max Alibaba closed n/a n/a n/a
2026-04-30 Grok 4.3 xAI closed 1M $1.25 / $2.50 n/a
2026-04-24 DeepSeek-V4 DeepSeek open 1M n/a FrontierCode 1.1 Main 17.6%
2026-04-23 GPT-5.5 OpenAI closed 1M $5 / $30 FrontierCode 1.1 Main 43%; Terminal-Bench 2.0 82.7%
2026-04-23 MiMo-V2.5-Pro Xiaomi open 1M n/a n/a
2026-04-20 Kimi K2.6 Moonshot AI open n/a n/a FrontierCode 1.0 Main 16%
2026-04-16 Claude Opus 4.7 Anthropic closed 1M $5 / $25 FrontierCode 1.1 Main 38.5%
2026-04-15 Qwen3.6 Alibaba open 131k n/a FrontierCode 1.0 Main 9.7%
2026-04-08 Muse Spark Meta closed n/a n/a Humanity's Last Exam 40.56% (Scale AI leaderboard); Terminal-Bench 4.0 14.55%
2026-04-07 GLM-5.1 Z.ai (Zhipu) open 200k n/a n/a
2026-04-02 Gemma 4 Google open 256k n/a MMLU-Pro 85.2% (31B); LiveCodeBench v6 80.0% (31B); AIME 2026 89.2% (31B, no tools)
2026-03-18 MiMo-V2-Pro Xiaomi closed n/a n/a n/a
2026-03-16 Mistral Small 4 Mistral AI open n/a n/a n/a
2026-03-11 Nemotron 3 Super NVIDIA open 1M n/a n/a
2026-03-10 Grok 4.20 xAI closed n/a n/a n/a
2026-03-05 GPT-5.4 OpenAI closed 1M $2.50 / $15 n/a
2026-03-05 GPT-5.4-mini OpenAI closed n/a n/a FrontierCode 1.1 Main 27%
2026-03-01 MiniMax M2.7 MiniMax open n/a n/a FrontierCode 1.0 Main 6%
2026-02-17 Claude Sonnet 4.6 Anthropic closed 1M $3 / $15 FrontierCode 1.1 Main 24.3%
2026-02-16 Qwen3.5 Alibaba open 131k n/a n/a
2026-02-14 Doubao Seed 2.0 ByteDance closed n/a n/a n/a
2026-02-11 GLM-5 Z.ai (Zhipu) open 200k n/a GPQA Diamond 86.0%
2026-02-05 Claude Opus 4.6 Anthropic closed 1M $5 / $25 FrontierCode 1.1 Main 26.6%; Humanity's Last Exam 34.44% (Scale AI leaderboard)
2026-02-01 Gemini 3.1 Flash Lite Google closed n/a n/a FrontierCode 1.0 Main 4.8%
2026-02-01 Gemini 3.1 Pro Google closed n/a n/a FrontierCode 1.0 Main 16.7%
2026-01-27 Kimi K2.5 Moonshot AI open n/a n/a FrontierCode 1.0 Main 6.9%
2026-01-15 MiniMax M2.5 MiniMax open n/a n/a FrontierCode 1.0 Main 5.3%
2025-12-23 MiniMax M2.1 MiniMax open n/a n/a n/a
2025-12-22 GLM-4.7 Z.ai (Zhipu) open n/a n/a n/a
2025-12-16 MiMo-V2-Flash Xiaomi open 256k n/a n/a
2025-12-15 Nemotron 3 NVIDIA open 1M n/a n/a
2025-12-11 GPT-5.2 OpenAI closed 400k $1.75 / $14 n/a
2025-12-02 Amazon Nova 2 Amazon closed 1M n/a n/a
2025-12-02 Mistral Large 3 Mistral AI open 256k $0.50 / $1.50 n/a
2025-12-01 DeepSeek-V3.2 DeepSeek open 128k $0.28 / $0.42 GPQA Diamond 82.4%
2025-11-24 Claude Opus 4.5 Anthropic closed 200k $5 / $25 n/a
2025-11-20 Olmo 3 AI2 open n/a n/a n/a
2025-11-18 Gemini 3 Pro Google closed 1M $2 / $12 GPQA Diamond 91.9%; Humanity’s Last Exam 37.5% (no tools)
2025-11-17 Grok 4.1 xAI closed 256k n/a n/a
2025-11-13 Ernie 5.0 Baidu closed n/a n/a n/a
2025-11-12 GPT-5.1 OpenAI closed 400k $1.25 / $10 n/a
2025-11-06 Kimi K2 Thinking Moonshot AI open 256k n/a n/a
2025-10-22 MiniMax-M2 MiniMax open 205k $0.30 / $1.20 n/a
2025-10-15 Claude Haiku 4.5 Anthropic closed 200k $1 / $5 Terminal-Bench 41.75% (32K thinking)
2025-10-09 Ling-1T Ant Group open n/a n/a n/a
2025-10-02 Granite 4.0 IBM open n/a n/a n/a
2025-09-30 GLM-4.6 Z.ai (Zhipu) open 200k n/a n/a
2025-09-29 Claude Sonnet 4.5 Anthropic closed 200k $3 / $15 OSWorld 61.4%
2025-09-23 Qwen3-Max Alibaba closed 262k n/a n/a
2025-08-29 LongCat-Flash-Chat Meituan open n/a n/a n/a
2025-08-28 MAI-1-preview Microsoft closed n/a n/a n/a
2025-08-21 DeepSeek-V3.1 DeepSeek open 128k n/a n/a
2025-08-20 Seed-OSS-36B ByteDance open n/a n/a n/a
2025-08-18 Nemotron Nano 2 NVIDIA open 131k n/a n/a
2025-08-07 GPT-5 OpenAI closed 400k $1.25 / $10 GPQA Diamond 88.4% (GPT-5 pro)
2025-08-05 Claude Opus 4.1 Anthropic closed 200k $15 / $75 n/a
2025-08-05 gpt-oss-120b OpenAI open 128k n/a GPQA Diamond 80.1%
2025-07-28 GLM-4.5 Z.ai (Zhipu) open 128k $0.60 / $2.20 GPQA 79.1%
2025-07-11 Kimi K2 Moonshot AI open 128k $0.60 / $2.50 LiveCodeBench v6 53.7%; AIME 2025 49.5%
2025-07-09 Grok 4 xAI closed 256k $3 / $15 Humanity’s Last Exam 25.4% (no tools); GPQA 87.5%
2025-06-27 Hunyuan-A13B Tencent open n/a n/a n/a
2025-06-16 MiniMax-M1 MiniMax open 1M n/a n/a
2025-06-10 Magistral Mistral AI open n/a n/a n/a
2025-05-28 DeepSeek-R1-0528 DeepSeek open 128k n/a n/a
2025-05-22 Claude 4 (Opus 4 / Sonnet 4) Anthropic closed 200k $15 / $75 Terminal-bench 43.2% (Opus 4); GPQA Diamond 79.6% (Opus 4)
2025-05-07 Mistral Medium 3 Mistral AI closed n/a n/a n/a
2025-04-30 Amazon Nova Premier Amazon closed n/a n/a n/a
2025-04-28 Qwen3 Alibaba open 131k n/a AIME 2024 85.7 (235B-A22B); AIME 2025 81.5 (235B-A22B); LiveCodeBench 70.7 (235B-A22B)
2025-04-16 o3 OpenAI closed 200k $10 / $40 GPQA Diamond 83.3%
2025-04-14 GPT-4.1 OpenAI closed 1M $2 / $8 GPQA Diamond 66.3%
2025-04-07 Llama-3.1-Nemotron-Ultra-253B NVIDIA open 131k n/a GPQA (reasoning on) 76.0%
2025-04-05 Llama 4 (Scout/Maverick) Meta open 10M n/a GPQA Diamond 69.8% (Maverick)
2025-03-25 Gemini 2.5 Pro Google closed 1M $1.25 / $10 GPQA Diamond 84.0%
2025-03-16 Ernie 4.5 Baidu open n/a n/a n/a
2025-03-12 Gemma 3 Google open 128k n/a GPQA Diamond 42.4% (27B)
2025-02-27 GPT-4.5 OpenAI closed 128k $75 / $150 GPQA Diamond 71.4%
2025-02-24 Claude 3.7 Sonnet Anthropic closed 200k $3 / $15 GPQA Diamond 78.2% (extended thinking)
2025-02-17 Grok 3 xAI closed 131k $3 / $15 GPQA Diamond 84.6% (Think)
2025-01-22 Doubao 1.5 Pro ByteDance closed n/a n/a n/a
2025-01-20 DeepSeek-R1 DeepSeek open 128k $0.55 / $2.19 GPQA Diamond 71.5%
2024-12-26 DeepSeek-V3 DeepSeek open 128k $0.27 / $1.10 GPQA Diamond 59.1%
2024-12-11 Gemini 2.0 Flash Google closed 1M n/a GPQA Diamond 62.1%
2024-12-03 Amazon Nova Amazon closed 300k n/a n/a
2024-11-05 Hunyuan-Large Tencent open 256k n/a n/a
2024-09-19 Qwen2.5 Alibaba open 131k n/a MMLU 85+; HumanEval 85+; MATH 80+
2024-09-12 o1 OpenAI closed 128k $15 / $60 GPQA Diamond 78.0%
2024-08-13 Grok-2 xAI open n/a n/a n/a
2024-07-23 Llama 3.1 405B Meta open 128k n/a MMLU 88.6%; HumanEval 89.0%; GSM8K 96.8%
2024-06-20 Claude 3.5 Sonnet Anthropic closed 200k $3 / $15 GPQA Diamond 59.4%
2024-06-14 Nemotron-4 340B NVIDIA open 4k n/a Arena Hard 54.2; MT-Bench (GPT-4-Turbo judge) 8.22
2024-05-13 GPT-4o OpenAI closed 128k $5 / $15 GPQA 53.6%
2024-05-06 DeepSeek-V2 DeepSeek open 128k $0.14 / $0.28 MMLU 77.8 (Chat RL); HumanEval 81.1 (Chat RL); GSM8K 92.2 (Chat RL)
2024-04-18 Llama 3 (8B/70B) Meta open 8k n/a MMLU 82.0% (70B); HumanEval 81.7% (70B); GSM8K 93.0% (70B)
2024-03-04 Claude 3 (Haiku/Sonnet/Opus) Anthropic closed 200k $15 / $75 GPQA Diamond 50.4% (Opus)
2024-02-26 Mistral Large Mistral AI closed 32k $8 / $24 MMLU 81.2%; HumanEval 45.1%
2024-02-15 Gemini 1.5 Pro Google closed 1M n/a MMLU 81.9%; GSM8K 91.7%; HumanEval 71.9%
2023-12-06 Gemini 1.0 (Ultra/Pro/Nano) Google closed 32k n/a MMLU 90.0% (Ultra, CoT@32); GSM8K 94.4% (Ultra); HumanEval 74.4% (Ultra)
2023-11-03 Grok-1 xAI open 8k n/a MMLU 73.0%; GSM8K 62.9%; HumanEval 63.2%
2023-09-27 Mistral 7B Mistral AI open 8k n/a MMLU 60.1%; HumanEval 30.5%; GSM8K 52.2% (8-shot)
2023-07-18 Llama 2 Meta open 4k n/a MMLU 68.9% (70B); GSM8K 56.8% (70B); HumanEval 29.9% (70B)
2023-07-11 Claude 2 Anthropic closed 100k $11.02 / $32.68 HumanEval (Codex P@1) 71.2%; GSM8K 88.0%; Bar exam (MBE) 76.5%
2023-05-10 PaLM 2 Google closed 8k n/a n/a
2023-03-16 Ernie Bot Baidu closed n/a n/a n/a
2023-03-14 Claude 1 Anthropic closed 9k n/a n/a
2023-03-14 GPT-4 OpenAI closed 8k $30 / $60 MMLU 86.4%; HumanEval 67.0%; GSM8K 92.0%
2022-11-30 GPT-3.5 / ChatGPT OpenAI closed 4k n/a n/a

Table 5. Launch date, weights, context, list price and vendor launch scores, newest first. Source: vendor announcements.

Closing

Frontier launch scores keep climbing on coding, agent, and knowledge evals, even when those tests get harder.

Open-weight models keep posting useful scores at lower list prices per million tokens than closed peers. That cost gap matters when work has to run at scale.

Those same models are also the main price pressure on list-price charts. Published parameter counts now land in the same ballpark as the closed models that disclose size, and that size overlap is one factor that helps explain why open models look closer on capability. Whether open-weight keeps closing on the closed frontier, and how fast, is still unsettled, though the recent record points that way.

Sources

These groups list the sources for scores, milestones, prices, open-model sizes and release dates. The compiled table and original charts may be reused with attribution to ryansmedstad.com under a Creative Commons Attribution 4.0 license. Formatted citation and BibTeX are in the Copy menu at the top of this page.

Intelligence and capability 7 sources

Figure 1 coding scores and Figure 3 capability axis: Cognition FrontierCode leaderboard, Cognition FrontierCode 1.1 announcement. Figure 1 agent scores: Terminal-Bench 4.0 leaderboard. Figure 1 Gemini 4 Argon vendor-reported Terminal-Bench score: Google DeepMind Gemini 4 Argon evals methodology. Figure 1 knowledge scores: Artificial Analysis Humanity's Last Exam leaderboard. Table 1 vendor-reported Humanity's Last Exam scores: Anthropic newsroom. Figure 1 work scores: Artificial Analysis GDPval-AA leaderboard.

Milestones 10 sources

Figure 2 Claude Fable 5 export-control suspension: Anthropic post on redeploying Claude Fable 5. Figure 2 decision-model milestone: TypeSafe AI post introducing Jev, Cloudflare Clef on Workers AI changelog, OpenAI Decisions API guide, Strands Decider announcement, Ollama decision-model support post. Figure 2 agent source-control milestone: Cloudflare Artifacts announcement, GitLab next-gen SCM post, GitHub post on agent-scale Git infrastructure, Anthropic post on Scion source control.

Pricing 11 sources

Figure 3 list prices: OpenAI API pricing, Anthropic API pricing, Google DeepMind API pricing, Meta AI (Superintelligence Labs) API pricing, Mistral AI API pricing, DeepSeek API pricing, xAI API pricing, Alibaba (Qwen) API pricing, Moonshot AI API pricing, Z.ai (Zhipu) API pricing, Thinking Machines Lab API pricing.

Open models and parameters 7 sources

Figure 4 open-model coverage: Hugging Face State of Open Models: Summer 2026. Figure 5 Llama 4 Maverick size: Meta Llama 4 announcement. Figure 5 Llama 4 Scout size: Llama 4 Scout model card. Figure 5 Nemotron 3.5 Bolt size: Nemotron 3.5 Bolt 74B-A9B model card. Figure 5 open-weight model cards: Meta models on Hugging Face, NVIDIA models on Hugging Face, Xiaomi MiMo models on Hugging Face.

Release dates 12 sources

Table 5 launch dates, context windows, and status: OpenAI models page, Anthropic models overview, Google Gemini models page, Meta Muse models page, xAI models page, Mistral models page, DeepSeek API news, NVIDIA Nemotron hub, Qwen blog, Moonshot AI platform docs, Z.ai docs, Thinking Machines blog.