Rank int64 1 22 | Model Name stringlengths 7 36 | Provider stringlengths 4 30 | Intelligence Index float64 12 53 | SWE-bench Verified stringlengths 5 5 | LiveCodeBench stringlengths 5 5 | Terminal-Bench 2.0 stringlengths 5 5 | Context Window stringclasses 6
values | Price (Input / 1M) float64 | Price (Output / 1M) float64 | Speed (Tokens/sec) int64 37 358 | Model Page URL stringlengths 46 67 |
|---|---|---|---|---|---|---|---|---|---|---|---|
1 | Claude Fable 5.1 (max with fallback) | Anthropic | 53 | 54.6% | 50.0% | 52.0% | 1,000,000 | null | null | 68 | https://vibecoderjournal.com/rankings/claude-fable-5-1-max/ |
2 | GPT-6 Astra (max) | OpenAI | 53 | 55.0% | 50.5% | 51.5% | 1,000,000 | null | null | 57 | https://vibecoderjournal.com/rankings/gpt-6-astra-max/ |
3 | Claude Opus 5 (max) | Anthropic | 51 | 52.6% | 48.0% | 50.0% | 1,000,000 | null | null | 51 | https://vibecoderjournal.com/rankings/claude-opus-5-max/ |
4 | GPT-5.6 Sol (max) | OpenAI | 47 | 48.0% | 44.5% | 46.5% | 1,000,000 | null | null | 61 | https://vibecoderjournal.com/rankings/gpt-5-6-sol-max/ |
5 | Qwen3.8 Max (0902) | Alibaba | 45 | 46.0% | 42.0% | 44.5% | 984,000 | null | null | 37 | https://vibecoderjournal.com/rankings/qwen-3-8-max-0902/ |
6 | Kimi K3 (max) | Kimi | 44 | 44.8% | 41.5% | 43.8% | 1,050,000 | null | null | 38 | https://vibecoderjournal.com/rankings/kimi-k3-max/ |
7 | Step 5 Preview | StepFun | 44 | 44.6% | 41.2% | 44.0% | 1,000,000 | null | null | 100 | https://vibecoderjournal.com/rankings/step-5-preview/ |
8 | GPT-5.6 Terra (max) | OpenAI | 42 | 43.0% | 39.5% | 41.5% | 1,000,000 | null | null | 84 | https://vibecoderjournal.com/rankings/gpt-5-6-terra-max/ |
9 | Gemini 3.8 Flash (high) | Google | 41 | 41.8% | 38.5% | 40.5% | 1,000,000 | null | null | 298 | https://vibecoderjournal.com/rankings/gemini-3-8-flash-high/ |
10 | Qwen3.8-Flash-Next | Alibaba | 40 | 40.6% | 38.0% | 40.0% | 256,000 | null | null | 55 | https://vibecoderjournal.com/rankings/qwen-3-8-flash-next/ |
11 | Claude Sonnet 5 (max) | Anthropic | 38 | 38.6% | 36.0% | 38.0% | 1,000,000 | null | null | 74 | https://vibecoderjournal.com/rankings/claude-sonnet-5-max/ |
12 | GPT-5.6 Luna (max) | OpenAI | 38 | 38.8% | 35.5% | 37.5% | 1,000,000 | null | null | 133 | https://vibecoderjournal.com/rankings/gpt-5-6-luna-max/ |
13 | DeepSeek V4 Pro 0813 (max) | DeepSeek | 36 | 36.8% | 33.5% | 35.5% | 1,000,000 | null | null | 88 | https://vibecoderjournal.com/rankings/deepseek-v4-pro-0813-max/ |
14 | DeepSeek V4 Flash Vision (max) | DeepSeek | 35 | 35.6% | 33.0% | 34.5% | 1,000,000 | null | null | 214 | https://vibecoderjournal.com/rankings/deepseek-v4-flash-vision-max/ |
15 | K2 Horizon 375B A23B | Institute of Foundation Models | 31 | 31.8% | 28.5% | 30.5% | 524,000 | null | null | 80 | https://vibecoderjournal.com/rankings/k2-horizon-375b-a23b/ |
16 | Gemini 3.1 Pro Preview | Google | 30 | 30.2% | 27.5% | 30.2% | 1,000,000 | null | null | 119 | https://vibecoderjournal.com/rankings/gemini-3-1-pro-preview/ |
17 | MiniMax-M3 | MiniMax | 30 | 30.6% | 28.0% | 30.0% | 1,000,000 | null | null | 186 | https://vibecoderjournal.com/rankings/minimax-m3/ |
18 | Inkling | Thinking Machines | 26 | 26.6% | 24.5% | 25.5% | 1,000,000 | null | null | 81 | https://vibecoderjournal.com/rankings/inkling/ |
19 | Gemini 3.5 Flash-Lite | Google | 23 | 23.2% | 21.2% | 23.2% | 1,000,000 | null | null | 358 | https://vibecoderjournal.com/rankings/gemini-3-5-flash-lite/ |
20 | Muse Glimmer (high) | Meta | 18 | 18.6% | 16.5% | 17.5% | 131,000 | null | null | 93 | https://vibecoderjournal.com/rankings/muse-glimmer-high/ |
21 | Mistral Medium 3.5 | Mistral | 15 | 15.6% | 14.0% | 14.5% | 256,000 | null | null | 137 | https://vibecoderjournal.com/rankings/mistral-medium-3-5/ |
22 | gpt-oss-120b (high) | OpenAI | 12 | 12.6% | 11.5% | 11.5% | 131,000 | null | null | 181 | https://vibecoderjournal.com/rankings/gpt-oss-120b-high/ |
license: mit task_categories: - text-generation - code-generation tags: - benchmarks - llm - swe-bench - livecodebench - evaluation size_categories: - n<1K
Frontier LLM Benchmarks & Coding Leaderboard (2026)
This dataset curates verified benchmark scores, context limits, pricing, and latency across 22 frontier AI models (including Claude Opus 5, GPT-6 Astra, DeepSeek, Gemini, and Mistral).
📊 Full Interactive Leaderboard & Methodology
For the live interactive rankings, real-time token calculators, and in-depth model cards, visit: 👉 VibeCoderJournal AI Rankings
Key Benchmarks Included
- SWE-bench Verified: Real-world GitHub software engineering issue resolution.
- LiveCodeBench: Contamination-free fresh code generation.
- Terminal-Bench 2.0: Autonomous multi-step CLI and shell agents.
- Context Windows & Speeds: Measured token throughput and context limits.
Source & Attribution
Data collected, cross-verified, and published by VibeCoderJournal.
- Downloads last month
- 26