Reasoning
ForecastBench Leaderboard
ForecastBench, from the Forecasting Research Institute, asks models to predict real-world future events, with questions that only resolve after the model's training cutoff so the answers can't have been memorised. It is scored as a difficulty-adjusted Brier index, where higher is better.
Source: epoch24 open models ranked+71 proprietaryData through Sep 2026
All models ranked on ForecastBench
Proprietary / closed models are shown dimmed — you can't run them locally, but they show where the open field stands.
| # | Model | Score | VRAM @ Q4 |
|---|---|---|---|
| 1 | GPT 5.5 (high) · proprietary | 66.7 | API only |
| 2 | Claude Opus 5 Max · proprietary | 66.6 | API only |
| 3 | GPT 6 Astra (high) · proprietary | 66.6 | API only |
| 4 | Gemini 3.7 Flash (high) · proprietary | 66.2 | API only |
| 5 | Claude Opus 4.8 Max · proprietary | 66.1 | API only |
| 6 | GPT 5.6 Sol (medium) · proprietary | 66.0 | API only |
| 7 | Claude Fable 5 Max · proprietary | 65.9 | API only |
| 8 | GPT 5.6 Sol Promax · proprietary | 65.8 | API only |
| 9 | Grok 4.5 (high) · proprietary | 65.8 | API only |
| 10 | Claude Fable 5 (high) · proprietary | 65.7 | API only |
| 11 | Claude Fable 5.1 (medium) · proprietary | 65.7 | API only |
| 12 | Gemini 3.6 Flash (medium) · proprietary | 65.7 | API only |
| 13 | GPT 5.6 Sol Max · proprietary | 65.7 | API only |
| 14 | Muse Spark 1.3 Max · proprietary | 65.5 | API only |
| 15 | Grok 4.6 (xhigh) · proprietary | 65.4 | API only |
| 16 | Gemini 3.1 Pro Preview (high) · proprietary | 65.2 | API only |
| 17 | Gemini 3 Pro Preview · proprietary | 65.0 | API only |
| 18 | GPT 5 (Aug 07, 2025, unspecified) · proprietary | 65.0 | API only |
| 19 | Grok 4 (Jul 09) · proprietary | 64.7 | API only |
| 20 | Claude Opus 4.5 (Nov 01, 2025, unspecified) · proprietary | 64.6 | API only |
| 21 | Grok 4 Fast · proprietary | 64.6 | API only |
| 22 | Grok 4.1 Fast Reasoning · proprietary | 64.6 | API only |
| 23 | GPT 4.5 Preview (Feb 27, 2025) · proprietary | 64.5 | API only |
| 24 | Claude Opus 4.1 (Aug 05, 2025, unspecified) · proprietary | 64.4 | API only |
| 25 | Gemini 3 Flash Preview · proprietary | 64.4 | API only |
| 26 | O3 (Apr 16, 2025, unspecified) · proprietary | 64.3 | API only |
| 27 | Gemini 2.5 Pro Preview (Mar 25) · proprietary | 64.2 | API only |
| 28 | GPT 5.5 (unspecified) · proprietary | 64.2 | API only |
| 29 | Kimi K3 · 2779.9B | 64.2 | ~1674 GB |
| 30 | Gemini 3.1 Pro Preview · proprietary | 64.1 | API only |
| 31 | Grok 4.20 0309 Reasoning · proprietary | 64.1 | API only |
| 32 | Claude Sonnet 4.6 (unspecified) · proprietary | 64.0 | API only |
| 33 | Gemini 2.5 Pro · proprietary | 64.0 | API only |
| 34 | Gemini 3.5 Flash (unspecified) · proprietary | 64.0 | API only |
| 35 | MiniMax M3 · 427.0B | 64.0 | ~257 GB |
| 36 | Claude Sonnet 4 (May 14, 2025, unspecified) · proprietary | 63.9 | API only |
| 37 | Claude Sonnet 4.5 (Sep 29, 2025, unspecified) · proprietary | 63.9 | API only |
| 38 | Claude Sonnet 4.6 (16K) · proprietary | 63.9 | API only |
| 39 | O4 Mini (Apr 16, 2025, unspecified) · proprietary | 63.9 | API only |
| 40 | Qwen3 235B A22B · 235.1B | 63.9 | ~142 GB |
| 41 | GPT 5.2 (Dec 11, 2025, unspecified) · proprietary | 63.8 | API only |
| 42 | Claude 3.7 Sonnet (Feb 19, 2025, unspecified) · proprietary | 63.7 | API only |
| 43 | O3 Mini (Jan 31, 2025, unspecified) · proprietary | 63.5 | API only |
| 44 | GLM 5 · 753.9B | 63.4 | ~457 GB |
| 45 | Grok 4.3 (unspecified) · proprietary | 63.2 | API only |
| 46 | Claude Haiku 4.5 (Oct 01, 2025, unspecified) · proprietary | 63.1 | API only |
| 47 | Claude Opus 4.6 (unspecified) · proprietary | 63.1 | API only |
| 48 | Claude Sonnet 5 (16K) · proprietary | 63.1 | API only |
| 49 | Grok 4.20 · proprietary | 63.1 | API only |
| 50 | GPT 5 Mini (Aug 07, 2025, unspecified) · proprietary | 63.0 | API only |
| 51 | GPT 5.4 (Mar 05, 2026, unspecified) · proprietary | 63.0 | API only |
| 52 | Claude 3.5 Sonnet (Jun 20, 2024) · proprietary | 62.9 | API only |
| 53 | Claude 3.5 Sonnet (Oct 22, 2024) · proprietary | 62.9 | API only |
| 54 | GPT 4 Turbo (Apr 09, 2024) · proprietary | 62.9 | API only |
| 55 | Claude Opus 4.7 (unspecified) · proprietary | 62.8 | API only |
| 56 | Gemini 2.5 Flash · proprietary | 62.7 | API only |
| 57 | Gemini 3.1 Flash Lite · proprietary | 62.6 | API only |
| 58 | Claude Opus 4.8 (24K) · proprietary | 62.4 | API only |
| 59 | DeepSeek v3 · 684.5B | 62.4 | ~415 GB |
| 60 | Gemini 2.5 Flash Preview (Apr 17) · proprietary | 62.4 | API only |
| 61 | Kimi K2 Instruct 0905 · 1026.4B | 62.4 | ~620 GB |
| 62 | Llama 3.1 405B Instruct · 405.9B | 62.3 | ~268 GB |
| 63 | Claude 3 Opus (Feb 29, 2024) · proprietary | 62.1 | API only |
| 64 | GPT 4.1 (Apr 14, 2025) · proprietary | 62.0 | API only |
| 65 | GPT 5.1 (Nov 13, 2025, unspecified) · proprietary | 62.0 | API only |
| 66 | Kimi K2 Instruct · 1026.4B | 62.0 | ~620 GB |
| 67 | Qwen3 235B A22B Thinking 2507 · 235.1B | 62.0 | ~142 GB |
| 68 | GLM 4.5 Air · 110.5B | 61.7 | ~67 GB |
| 69 | Claude Opus 4.8 (unspecified) · proprietary | 61.6 | API only |
| 70 | Mistral Large Instruct 2407 · 122.6B | 61.5 | ~81 GB |
| 71 | Qwen2.5 72B Instruct · 72.7B | 61.4 | ~45 GB |
| 72 | Mistral Large Instruct 2411 · 122.6B | 61.3 | ~75 GB |
| 73 | GPT 4 (Jun 13) · proprietary | 61.2 | API only |
| 74 | Gemma 4 31B IT · 31.3B | 61.1 | ~20 GB |
| 75 | GPT 4o (May 13, 2024) · proprietary | 61.0 | API only |
| 76 | Llama 3.3 70B Instruct · 70.6B | 61.0 | ~47 GB |
| 77 | GPT 5.4 Mini (Mar 17, 2026, unspecified) · proprietary | 60.8 | API only |
| 78 | DeepSeek V4 Pro · 1598.8B | 60.7 | ~960 GB |
| 79 | DeepSeek V3.1 · 684.5B | 60.5 | ~415 GB |
| 80 | GPT 4o (Nov 20, 2024) · proprietary | 60.4 | API only |
| 81 | Meta Llama 3 70B · 70.6B | 59.9 | ~47 GB |
| 82 | Qwen1.5 110B Chat · 111.2B | 59.3 | ~68 GB |
| 83 | Meta Llama 3 8B Instruct · 8.0B | 59.1 | ~5.3 GB |
| 84 | Llama 4 Scout 17B 16E Instruct · 108.6B | 58.9 | ~72 GB |
| 85 | GPT 5 Nano (Aug 07, 2025, unspecified) · proprietary | 58.7 | API only |
| 86 | Gemini 1.5 Pro 001 · proprietary | 58.6 | API only |
| 87 | GPT 5.4 Nano (Mar 17, 2026, unspecified) · proprietary | 58.6 | API only |
| 88 | Mixtral 8x22B Instruct v0.1 · 140.6B | 58.6 | ~85 GB |
| 89 | Claude 2.1 · proprietary | 58.5 | API only |
| 90 | Gemini 1.5 Flash 001 · proprietary | 58.4 | API only |
| 91 | Mixtral 8x7B Instruct v0.1 · 46.7B | 58.0 | ~29 GB |
| 92 | Gemini 2.0 Flash Lite 001 · proprietary | 56.8 | API only |
| 93 | Claude 3 Haiku (Mar 07, 2024) · proprietary | 54.6 | API only |
| 94 | Llama 2 70B Chat HF · 69.0B | 50.2 | ~152 GB |
| 95 | GPT 3.5 Turbo (Jan 25) · proprietary | 46.2 | API only |
Score vs model size
Which models give the most quality for their size — the ones worth running locally.
- Meta Llama 3 8B Instruct, 8B, score 59.1 — on the efficiency frontier (best score at its size or smaller).
- Gemma 4 31B IT, 31B, score 61.1 — on the efficiency frontier (best score at its size or smaller).
- Qwen2.5 72B Instruct, 73B, score 61.4 — on the efficiency frontier (best score at its size or smaller).
- GLM 4.5 Air, 110B, score 61.7 — on the efficiency frontier (best score at its size or smaller).
- Qwen3 235B A22B, 235B, score 63.9 — on the efficiency frontier (best score at its size or smaller).
- MiniMax M3, 427B, score 64.0 — on the efficiency frontier (best score at its size or smaller).
- Kimi K3, 2.8T, score 64.2 — on the efficiency frontier (best score at its size or smaller).
ForecastBench: frequently asked questions
- What is the best open LLM on ForecastBench?
- Kimi K3 is the top open model on ForecastBench, scoring 64.2. Among all models tested — including proprietary ones — it ranks #27. The top model overall is GPT 5.5 (high) (OpenAI) at 66.7.
- What's the best ForecastBench model you can run on a 24 GB GPU?
- Gemma 4 31B IT is the highest-scoring open model that fits in 24 GB at 4-bit quantization (~20 GB), scoring 61.1 on ForecastBench.
- What's the best ForecastBench model you can run on a 12 GB GPU?
- Meta Llama 3 8B Instruct is the highest-scoring open model that fits in 12 GB at 4-bit quantization (~5.3 GB), scoring 59.1 on ForecastBench.
- Can open models match proprietary models on ForecastBench?
- Not quite on ForecastBench: the strongest proprietary model (GPT 5.5 (high)) scores 66.7, ahead of the best open model (Kimi K3) at 64.2 — but you can run the open one yourself.
Scores aggregated from epoch. llmrun does not run this benchmark — see the source for methodology, or the about benchmarks for what it measures.