Open LLM Leaderboard

Name: Open LLM Leaderboard - LLM Stats
Creator: LLM Stats
License: https://llm-stats.com/legal/terms-of-service

Ranking the best open LLMs by performance, price, and speed

Select 2-4 models to compare

← Scroll →


GLM-5.1	🇨🇳	Open	200k	$1.40	$4.40	229c/s	1,754	54.2	46.8	43.0	29.7	–	39.0	28.9	–	–	–	–	86.2%	—	—	—	—	—	79.3%	—	—	—	71.8%	52.3%	—	—	40.7%	—	—	—	—	—	—	58.4%	754	16.0s	-	Apr. 2026	ZAI
GLM-5	🇨🇳	Open	200k	$1.00	$3.20	336c/s	1,595	51.5	–	36.1	26.3	–	–	25.6	–	–	–	–	—	—	77.8%	—	—	—	75.9%	—	—	—	67.8%	—	—	—	—	—	—	—	—	—	—	—	744	7.4s	-	Feb. 2026	ZAI
Kimi K2.6	🇨🇳	Open	262.1k	$0.95	$4.00	46c/s	1,562	58.1	49.7	43.7	37.3	–	37.1	31.3	–	–	–	–	90.5%	—	80.2%	—	—	—	86.3%	86.7%	80.1%	—	—	36.4%	—	—	50.0%	—	—	—	—	52.2%	27.9%	58.6%	1000	60.7s	-	Apr. 2026	MoonshotAI
Kimi K2.5	🇨🇳	Open	-	-	-	–	1,462	49.8	47.5	31.4	30.0	–	35.3	13.7	38.2	46.8	46.8	48.0	87.6%	96.1%	76.8%	—	—	—	74.9%	77.5%	78.5%	—	—	50.2%	—	—	—	—	—	—	—	48.7%	—	50.7%	1000	–	-	Jan. 2026	MoonshotAI
Qwen3.5-397B-A17B	🇨🇳	Open	262.1k	$0.60	$3.60	105c/s	1,294	48.9	46.0	30.0	25.7	28.7	27.8	21.9	38.7	53.1	53.2	52.6	88.4%	—	76.4%	—	88.5%	—	69.0%	—	—	—	—	28.7%	—	—	38.3%	—	—	—	—	—	—	—	397	19.2s	-	Feb. 2026	Qwen
DeepSeek-V4-Pro-Max	🇨🇳	Open	1.0M	$1.74	$3.48	37c/s	1,280	57.0	52.9	43.5	32.9	–	33.5	33.1	19.8	42.0	42.5	47.8	90.1%	—	80.6%	—	—	—	83.4%	—	—	—	73.6%	48.2%	57.9%	—	51.8%	—	—	—	—	—	—	55.4%	1600	149.0s	-	Apr. 2026	DeepSeek
Gemma 4 31B	🇺🇸	Open	262.1k	$0.14	$0.40	216c/s	1,258	44.8	39.5	–	–	–	27.1	19.4	29.9	41.9	41.9	36.3	84.3%	—	—	—	88.4%	—	—	—	76.9%	—	—	26.5%	—	—	—	—	—	—	—	—	—	—	30.7	860ms	Jan. 2025	Apr. 2026	Google
Gemma 4 26B-A4B	🇺🇸	Open	262.1k	$0.13	$0.40	118c/s	1,241	35.5	32.8	–	–	–	20.5	17.8	15.8	32.3	32.3	30.6	82.3%	—	—	—	86.3%	—	—	—	73.8%	—	—	17.2%	—	—	—	—	—	—	—	—	—	—	25.2	2.7s	Jan. 2025	Apr. 2026	Google
MiniMax M2.7	🇨🇳	Open	204.8k	$0.30	$1.20	170c/s	1,156	53.1	–	38.9	–	–	–	23.1	–	27.8	28.4	–	—	—	—	—	—	—	—	—	—	—	—	—	—	—	46.3%	—	—	—	—	—	—	56.2%	-	4.0s	-	Mar. 2026	MiniMax
GLM-4.6	🇨🇳	Open	-	-	-	–	1,144	37.7	34.2	19.9	6.8	–	12.8	–	–	–	–	–	81.0%	93.9%	68.0%	—	—	—	45.1%	—	—	—	—	17.2%	—	—	—	40.5%	—	—	—	—	—	—	357	–	-	Sep. 2025	ZAI
DeepSeek-V4-Flash-Max	🇨🇳	Open	1.0M	$0.14	$0.28	72c/s	1,143	51.9	51.2	37.6	23.7	–	31.0	26.9	7.6	35.0	35.1	44.2	88.1%	—	79.0%	—	—	—	73.2%	—	—	—	69.0%	45.1%	34.1%	—	47.8%	—	—	—	—	—	—	52.6%	284	13.8s	-	Apr. 2026	DeepSeek
Qwen3.6-27B	🇨🇳	Open	262.1k	$0.60	$3.60	161c/s	1,118	45.7	42.8	34.3	–	–	30.2	23.4	29.1	46.1	46.1	45.9	87.8%	—	77.2%	—	—	82.9%	—	78.4%	75.8%	—	—	24.0%	—	—	—	—	—	—	—	—	—	53.5%	27.8	19.1s	-	Apr. 2026	Qwen
GLM-4.7	🇨🇳	Open	-	-	-	–	1,064	43.8	43.9	22.4	16.9	–	29.8	11.9	–	36.2	36.2	35.9	85.7%	95.7%	73.8%	—	—	—	52.0%	—	—	—	—	42.8%	—	—	—	33.3%	—	—	—	—	—	—	358	–	-	Dec. 2025	ZAI
MiniMax M2.5	🇨🇳	Open	1M	$0.30	$1.20	399c/s	1,056	52.5	–	38.2	26.8	–	–	–	–	-1.0	–	–	—	—	80.2%	—	—	—	76.3%	—	—	—	—	—	—	—	—	—	—	—	—	—	—	55.4%	230	1.7s	-	Feb. 2026	MiniMax
LongCat-Flash-Chat	🇨🇳	Open	128k	$0.30	$1.20	11c/s	1,039	23.4	22.7	15.7	–	17.6	–	14.0	–	34.7	34.7	34.2	73.2%	61.3%	60.4%	—	—	—	—	—	—	—	—	—	—	—	—	39.5%	—	—	—	—	—	—	560	3.8s	-	Aug. 2025	Meituan
MiniMax M2	🇨🇳	Open	1M	$0.30	$1.20	428c/s	971	33.6	25.6	23.3	4.8	19.0	6.1	13.9	–	29.6	29.5	29.3	78.0%	78.0%	69.4%	—	—	—	44.0%	—	—	—	—	12.5%	—	—	—	46.3%	—	—	—	36.0%	—	—	230	1.3s	-	Oct. 2025	MiniMax
DeepSeek-V3.2 (Non-thinking)	🇨🇳	Open	131.1k	$0.28	$0.42	288c/s	969	–	–	–	–	–	–	–	–	–	–	–	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	685	2.0s	-	Dec. 2025	DeepSeek
Kimi K2-Thinking-0905	🇨🇳	Open	-	-	-	–	941	45.2	47.0	25.9	19.8	-6.3	36.5	–	–	26.1	26.0	39.1	84.5%	100.0%	71.3%	—	—	—	60.2%	—	—	—	—	51.0%	—	—	—	47.1%	—	—	—	44.8%	—	—	1000	–	-	Sep. 2025	MoonshotAI
Mistral Large 3 (675B Instruct 2512)	🇫🇷	Open	262.1k	$0.50	$1.50	226c/s	867	10.2	19.7	-0.2	–	–	–	–	–	–	–	–	43.9%	—	—	—	85.5%	—	—	—	—	—	—	—	23.8%	—	—	—	—	—	—	—	—	—	675	1.7s	-	Dec. 2025	Mistral
MiniMax M2.1	🇨🇳	Open	1M	$0.30	$1.20	301c/s	856	41.3	36.2	26.9	19.1	18.2	17.9	20.3	19.9	49.9	49.9	49.7	81.0%	81.0%	67.0%	—	—	—	62.0%	—	—	—	—	22.0%	—	—	43.5%	47.9%	—	—	—	39.0%	—	—	230	1.3s	-	Dec. 2025	MiniMax
DeepSeek-V3.2-Speciale	🇨🇳	Open	-	-	-	–	832	43.2	45.8	22.6	–	–	24.5	12.6	–	–	–	–	—	96.0%	73.1%	—	—	—	—	—	—	—	—	30.6%	—	—	35.2%	—	—	—	—	—	—	—	685	–	-	Dec. 2025	DeepSeek
MiMo-V2-Flash	🇨🇳	Open	-	-	-	–	793	38.4	37.9	22.8	16.0	–	18.5	9.3	21.9	38.2	38.1	37.9	83.7%	94.1%	73.4%	—	—	—	58.3%	—	—	—	—	22.1%	—	—	—	30.5%	—	—	—	—	—	—	309	–	-	Dec. 2025	Xiaomi
LongCat-Flash-Thinking	🇨🇳	Open	-	-	-	–	791	35.9	34.0	18.9	–	23.7	16.1	21.2	–	36.3	32.1	31.9	81.5%	90.6%	59.4%	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	560	–	-	Sep. 2025	Meituan
Qwen3.5-27B	🇨🇳	Open	262.1k	$0.30	$2.40	103c/s	775	42.0	42.4	21.4	20.4	22.2	29.4	12.7	30.9	45.6	45.7	43.9	85.5%	—	72.4%	—	85.9%	82.3%	61.0%	79.5%	75.0%	70.3%	—	48.5%	—	—	—	—	—	—	—	—	—	—	27	36.0s	-	Feb. 2026	Qwen
GLM-4.7-Flash	🇨🇳	Open	-	-	-	–	759	31.4	29.1	8.4	3.4	–	7.4	9.6	–	–	–	–	75.2%	91.6%	59.2%	—	—	—	42.8%	—	—	—	—	14.4%	—	—	—	—	—	—	—	—	—	—	30	–	-	Jan. 2026	ZAI
LongCat-Flash-Lite	🇨🇳	Open	256k	$0.10	$0.40	64c/s	756	23.0	19.9	11.7	–	18.6	–	14.9	–	22.0	22.1	21.7	66.8%	63.2%	54.4%	—	—	—	—	—	—	—	—	—	—	—	—	33.8%	—	—	—	—	—	—	68.5	3.5s	-	Feb. 2026	Meituan
DeepSeek-V3.2-Exp	🇨🇳	Open	-	-	-	–	750	35.7	33.4	21.7	1.5	–	15.8	–	–	38.7	38.7	38.4	79.9%	89.3%	67.8%	—	—	—	40.1%	—	—	—	—	19.8%	97.1%	—	—	37.7%	—	—	—	—	—	—	685	–	-	Sep. 2025	DeepSeek
Qwen3.5-122B-A10B	🇨🇳	Open	262.1k	$0.40	$3.20	128c/s	749	43.0	44.0	25.8	22.1	24.2	30.7	13.4	32.3	48.7	48.8	47.4	86.6%	—	72.0%	—	86.7%	83.9%	63.8%	77.2%	76.9%	70.4%	—	47.5%	—	—	—	—	—	—	—	—	—	—	122	33.9s	-	Feb. 2026	Qwen
GLM-4.5	🇨🇳	Open	-	-	-	–	744	34.0	34.3	20.7	-4.6	25.1	8.2	25.6	–	41.6	37.5	37.3	79.1%	—	64.2%	—	—	—	26.4%	—	—	—	—	14.4%	—	—	—	37.5%	79.7%	—	—	41.7%	—	—	355	–	-	Jul. 2025	ZAI
Step-3.5-Flash	🇨🇳	Open	65.5k	$0.10	$0.40	363c/s	731	49.2	46.6	27.9	22.4	–	–	18.3	–	–	–	–	—	97.3%	74.4%	—	—	—	69.0%	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	196	8.8s	-	Feb. 2026	StepFun

Showing 1–30 of 180 models

Open LLM Leaderboard highlights

Independent ranking of open-weight large language models — Llama, Qwen, GLM, DeepSeek, Mistral, Kimi and more — by coding-arena score, GPQA Diamond, throughput, latency, and per-token pricing. Updated continuously from provider APIs and verified benchmarks. See the LLM Stats Score methodology for how rankings are computed.

Best for coding (Arena): Claude Opus 4.6 (21.3 arena score)
Best on GPQA Diamond: Claude Mythos Preview (94.6%)
Best on AIME 2025: Gemini 3 Pro (100.0%)
Best on SWE-Bench Verified: Claude Mythos Preview (93.9%)
Highest throughput: Mercury 2 (838 tok/s)
Lowest latency: Qwen3 30B A3B (297.86 s)
Cheapest input: Nemotron 3 Nano (30B A3B) ($0.06 / 1M tok)
Largest context window: Grok 4 Fast (2.0M tokens)

FAQ

Common questions about the open llm leaderboard

What is the best LLM right now?

Based on coding-arena performance — the most discriminating signal at the frontier — Claude Opus 4.6 currently leads. For knowledge-heavy reasoning (GPQA Diamond), Claude Mythos Preview scores highest. Choose by axis rather than a single ranking — see the highlights above for per-metric leaders.

How does the Open LLM Leaderboard rank models?

Models are sorted by coding-arena score (when available), then by GPQA Diamond. Each row aggregates verified benchmark results, provider-reported pricing, and live performance metrics (output throughput and time-to-first-token) sampled across the major API providers. See the LLM Stats Score methodology for the full weighting and refresh cadence.

How many models are tracked?

This leaderboard tracks 302 canonical models across every major lab and provider. New releases typically appear within hours.

Where does pricing data come from?

Per-model input/output pricing is pulled from each provider's public API price list and verified against billing samples from the LLM Stats proxy. When a model is hosted by multiple providers, the cheapest available rate is shown by default.

How is performance measured?

Output throughput (tokens/second) and time-to-first-token are measured by routing standardized prompts through each provider's API and averaging over a 7-day rolling window. Numbers update hourly. Per-model splits live on each model detail page.

How often does the data update?

Pricing and model metadata revalidate every hour. Live performance metrics update on a 7-day rolling average. Benchmark scores update when a new verified result is published or a new evaluation lands on LLM Stats.