Open-Source LLM Leaderboard
As of Oct 11, the top open-weight model on Arena is MiMo-V2.6-Pro.
Gemini 4 Argon1,533.9
Intelligence #1
Claude Opus 5.557.6
Coding #1Claude Fable 5.181.6
Best value
MiMo-V2.6-Flash$0.28/1M out
Open weights #1MiMo-V2.6-ProArena 1,491.7
FastestClaude Haiku 5.5237 tok/s
This week
vs Oct 4
Open-Source LLM Leaderboard: full table
Showing 23 rows| # | Model | Arena↓ | AA index | Price $/1M tokens | Latency (s) & speed | Context | Released | |||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Intel. | Coding | Input | Output | Lowest route | TTFT | Answer | tok/s | |||||
| 1 | MiMo-V2.6-ProXiaomiOpen weights | 1491.7±9.4 | 46.3 | – | $0.44 | $0.87 | $0.44 / $0.87 | 3.1 | 55 | 38 | 1.05M | Sep 21 |
| 2 |  Kimi K3Moonshot AIOpen weights | 1475.5±4.7 | 43.6 | 76.2 | $3.00 | $15.00 | $0.80 / $10.00 ⚠ | 2.9 | 54 | 39 | 1.05M | Jul 16 |
| 3 |  GLM-5.3Z.AIOpen weights | 1471.2±5.4 | 44.8 | 74.8 | $1.40 | $4.40 | $0.07 / $4.80 ⚠ | 3.2 | 36 | 60 | 1.05M | Aug 18 |
| 4 | GLM-5.3 FlashZ.AIOpen weights | 1471.1±5.0 | 41.8 | 71.5 | $0.15 | $0.50 | $0.15 / $0.50 | 2.6 | 38 | 56 | 1.05M | Aug 26 |
| 5 | GLM-5.2Z.AIOpen weights | 1469.8±4.3 | 33.7 | 68.8 | $1.40 | $4.40 | $0.06 / $7.00 ⚠ | – | – | – | 1.05M | Jun 16 |
| 6 | MiMo V2.5 ProXiaomiOpen weights | 1465.3±3.6 | 26.0 | 60.2 | $0.44 | $0.87 | $0.44 / $0.87 | – | – | – | 1.05M | Apr 22 |
| 7 |  DeepSeek V4.1 FlashDeepSeekOpen weights | 1462.4±6.5 | 39.5 | – | $0.30 | $1.20 | $0.30 / $1.20 | 0.8 | 10 | 215 | 1.05M | Sep 10 |
| 8 | GLM-5.1Z.AIOpen weights | 1461.4±3.6 | 26.1 | 55.8 | $1.29 | $4.07 | $0.97 / $3.04 | – | – | – | 205K | Apr 7 |
| 9 | Kimi K2.6Moonshot AIOpen weights | 1455.3±4.4 | 27.0 | 61.8 | $0.95 | $4.00 | $0.65 / $3.41 | – | – | – | 262K | Apr 20 |
| 10 | MiMo-V2.6-FlashXiaomiOpen weights | 1455.2±7.0 | 37.9 | – | $0.14 | $0.28 | $0.14 / $0.28 | 2.9 | 45 | 48 | 1.05M | Sep 21 |
| 11 | DeepSeek V4 ProDeepSeekOpen weights | 1451.1±3.9 | 36.0 | 68.8 | $1.32 | $3.96 | $0.21 / $0.42 ⚠ | 1.1 | 26 | 82 | 1.05M | Aug 13 |
| 12 | GLM-5Z AIOpen weights | 1446.2±4.3 | 27.9 | – | $1.00 | $3.20 | $0.60 / $1.92 | – | – | – | 205K | Feb 11 |
| 13 | Kimi K2.5Moonshot AIOpen weights | 1445.2±3.3 | 23.5 | 46.8 | $0.60 | $2.75 | $0.50 / $2.50 | – | – | – | 262K | Jan 27 |
| 14 |  Qwen3.8 27BAlibabaOpen weights | 1441.1±5.3 | 33.7 | 68.1 | $0.50 | $3.00 | $0.04 / $2.30 ⚠ | 1.2 | 45 | 46 | 1M | Aug 14 |
| 15 | GLM-4.6Z AIOpen weights | 1439.8±3.9 | 18.5 | 45.8 | $0.58 | $2.20 | $0.50 / $2.00 | – | – | – | 205K | Sep 30, 2025 |
| 16 |  Hy3TencentOpen weights | 1438.8±6.3 | 25.3 | 58.8 | $0.14 | $0.56 | $0.13 / $0.53 | 2.0 | 24 | 91 | 262K | Jul 6 |
| 17 |  MiniMax M3MiniMaxOpen weights | 1432.6±3.9 | 29.2 | 58.6 | $0.30 | $1.20 | $0.30 / $1.20 | 1.0 | 19 | 110 | 1.05M | Jun 1 |
| 18 | DeepSeek V4 FlashDeepSeekOpen weights | 1432.1±4.0 | 34.3 | 69.1 | $0.44 | $1.32 | $0.03 / $1.28 ⚠ | – | – | – | 1.05M | Jul 31 |
| 19 | MiMo V2.5XiaomiOpen weights | 1427.6±4.2 | 25.2 | 56.8 | $0.14 | $0.28 | $0.14 / $0.28 | – | – | – | 1.05M | Apr 22 |
| 20 | DeepSeek V3.2DeepSeekOpen weights | 1424.5±3.5 | 21.5 | 44.2 | $0.28 | $0.42 | $0.26 / $0.80 | – | – | – | 164K | Dec 1, 2025 |
| 21 |  Mistral Medium 3.5MistralOpen weights | 1420.7±6.4 | 14.2 | 46.9 | $1.50 | $7.50 | $1.50 / $7.50 | 0.6 | 13 | 160 | 262K | Apr 29 |
| 22 | Kimi K2.7 CodeMoonshot AIOpen weights | – | 25.8 | 60.8 | $0.95 | $4.00 | $0.67 / $3.35 | 1.1 | 39 | 59 | 262K | Jun 12 |
| 23 | Qwen3.8 2.4T A95BAlibabaOpen weights | – | 39.9 | 71.9 | $2.00 | $6.00 | $2.00 / $6.00 | 2.0 | 59 | 35 | 1.05M | Aug 12 |
Arena scores are LMArena's Bradley-Terry scale (not Elo). Input/output prices are official API list prices per million tokens; the lowest route price is the cheapest OpenRouter quote. In this view the median Arena margin is ±4.3 while neighbouring models sit a median 3.4 apart (19 of 20 adjacent pairs overlap): gaps under 4.3 say nothing about which model ranks higher.
One row per model: each row uses that model’s best tier for the column you sorted by (best of the tiers that have a measurement, ties going to the stronger tier). Arena scores and prices are model-level, so every tier of a model shares them. Switch to “Each tier” to see each reasoning tier on its own row.
Not yet ranked by Arena 2
Tracked models that have no LMArena or Artificial Analysis score yet, newest first. They move into the table once scored.
- GPT-6 Astra Pro OpenAI · 2026-09-04
- Seed 2.1 Turbo ByteDance · 2026-08-12
All 82 model pages
Claude Fable 5 · Claude Fable 5.1 · Claude Haiku 5.5 · Claude Opus 4.5 · Claude Opus 4.6 · Claude Opus 4.7 · Claude Opus 4.8 · Claude Opus 5 · Claude Opus 5.5 · Claude Sonnet 4.6 · Claude Sonnet 5 · Claude Sonnet 5.5 · DeepSeek V3.2 · DeepSeek V4 Flash · DeepSeek V4 Flash Vision · DeepSeek V4 Pro · DeepSeek V4 Pro 0424 · DeepSeek V4.1 Flash · Gemini 2.5 Pro · Gemini 3 Flash · Gemini 3 Pro · Gemini 3.1 Pro Preview · Gemini 3.5 Flash · Gemini 3.6 Flash · Gemini 3.7 Flash · Gemini 3.8 Flash · Gemini 4 Argon · GLM-4.6 · GLM-5 · GLM-5.1 · GLM-5.2 · GLM-5.3 · GLM-5.3 Flash · GPT-5.1 · GPT-5.2 · GPT-5.2 Chat · GPT-5.3 Codex · GPT-5.4 · GPT-5.5 · GPT-5.6 Luna · GPT-5.6 Sol · GPT-5.6 Terra · GPT-6 Astra · GPT-6 Astra Pro · GPT-6 Luna · GPT-6 Sol · GPT-6.1 Sol · Grok 4.5 · Grok 4.6 · Grok 4.7 · Grok 4.20 · Hy3 · JT-4.1 Flash 236B A21B · K2 Horizon 375B A23B · Kimi K2.5 · Kimi K2.6 · Kimi K2.7 Code · Kimi K3 · Ling 3.1 Flash · MiMo-V2.6-Flash · MiMo-V2.6-Pro · MiMo V2.5 · MiMo V2.5 Pro · MiniMax M3 · Mistral Large · Mistral Large 4 Preview · Mistral Medium 3.5 · Motif 3 · Muse Spark · Muse Spark 1.1 · Muse Spark 1.2 · Muse Spark 1.3 · Qwen3 Max · Qwen3.6 Max · Qwen3.7 Max · Qwen3.7 Plus · Qwen3.8 2.4T A95B · Qwen3.8 27B · Qwen3.8 Max · Qwen3.8-Flash-Next · Seed 2.1 Turbo · Step 5 Preview
Sources & methodology
As of Oct 11, 2026, the highest-rated open-weight model on Arena is MiMo-V2.6-Pro (1,491.7), followed by Kimi K3 (1,475.5) and GLM-5.3 (1,471.2). 21 open-weight models can be self-hosted; open weights do not mean unrestricted commercial use, so check each licence.
- LMArena — Arena scores, 95% confidence intervals, vote counts and ranks (text leaderboard; dataset under CC BY 4.0).
- Artificial Analysis — Intelligence Index, Coding Index, output speed, TTFT / first-answer latency, release dates and official prices (shown on this page only, not in the public API).
- OpenRouter — Per-provider prices (the cheapest becomes “lowest route”), context length and listing dates.
- Vendor pricing pages — “Input / Output” are official API list prices (USD per 1M tokens); when no official price exists we use the OpenRouter price and mark it with *.
How we compute it
- Ranking: by default, LMArena Arena score (a Bradley–Terry model, not Elo). The ± next to each score is the half-width of the 95% confidence interval, and neighbouring ranks often overlap. We do not invent a composite score.
- Intelligence / Coding Index: Artificial Analysis aggregates of public benchmarks. When a model has several reasoning-effort tiers, the table shows the best tier for the current sort column; switch to “by tier” to see each one.
- Prices are USD per 1M tokens. “Lowest route” is the cheapest OpenRouter provider’s input / output price; ⚠ marks routes more than 60% below list (usually promos or a single provider). The best-value board only includes models with Arena ≥ 1400 or Intelligence Index ≥ 40, sorted by official output price.
- Latency and speed: TTFT is the time to the first token (including thinking tokens); “Answer” is the time to the first answer token (includes the thinking phase for reasoning models); speed is output tokens per second. All are Artificial Analysis medians.
- Updates: synced automatically every day at 14:20 (UTC+8). LMArena publishes on its own schedule, so the “Arena published” date can be older than the update date. “This week” compares against the daily archive from 7 days earlier.
- Open data: /api/public/llm-leaderboard.php (JSON, LMArena and OpenRouter fields only).
FAQ
What is the best AI model right now?
As of Oct 11, 2026, Gemini 4 Argon leads the LMArena text leaderboard (1,533.9 ±8.9), ahead of Claude Opus 5.5 (1,514.8). On the Artificial Analysis Intelligence Index, Claude Opus 5.5 is first (57.6). The two measure different things — Arena is blind human preference, the index aggregates benchmarks — and gaps inside the confidence interval are not meaningful.
Which good AI model is the cheapest?
Among models above the quality floor (Arena ≥ 1400 or Intelligence Index ≥ 40), MiMo-V2.6-Flash has the lowest official price: $0.14 input / $0.28 output per 1M tokens, followed by MiMo V2.5 ($0.28 output). See the best-value leaderboard.
What is the best open-source (open-weight) AI model?
By Arena score, the top open-weight model is MiMo-V2.6-Pro (Xiaomi, 1,491.7), followed by Kimi K3 (1,475.5). We track 21 open-weight models with an Arena score — see the open-weight leaderboard. Open weights do not mean unrestricted commercial use; check each licence.
Which AI model is best for coding?
On the Artificial Analysis Coding Index, Claude Fable 5.1 is highest (81.6), followed by Claude Opus 5 (78.0). The index aggregates coding benchmarks and differs from the Arena score; see the full coding leaderboard.
Which AI model is the fastest?
The fastest measured output speed belongs to Claude Haiku 5.5 (237 tokens/s); the lowest time to first answer token belongs to DeepSeek V4.1 Flash (10.12 s). These are Artificial Analysis medians; for reasoning models the first-answer latency includes thinking time.
Which AI models were released recently?
6 tracked models were released in the last 14 days, including Claude Haiku 5.5 (Oct 7, 2026), Mistral Large 4 Preview (Oct 6, 2026), Ling 3.1 Flash (Oct 1, 2026), Gemini 4 Argon (Sep 30, 2026). New models usually take a few days to appear on LMArena; until then they are listed under “Not yet ranked on Arena”.
How often is the data updated, and where does it come from?
Data syncs automatically every day at 14:20 (UTC+8). Arena scores come from LMArena (current leaderboard published Oct 8, 2026); Intelligence and Coding Index, speed and latency from Artificial Analysis; lowest-route prices and context length from OpenRouter; official prices from Artificial Analysis and vendor pricing pages. Last update: Oct 11, 2026.
Why do prices differ from the official site?
Input/output prices are official list prices in USD per million tokens. The lowest route price is the cheapest OpenRouter route, which may be a promotion or a single provider and can differ from local resellers.