Leaderboards
The same models, ranked by whichever question you are actually asking. Every board is rebuilt daily from the latest data. For the overall ranking, see the LLM leaderboard 2026.
Got two in mind already? Put any models side by side.
This month’s picks: Best LLM overall (October 2026) · Best LLM for coding (October 2026) · Best LLM for agents (October 2026) · Best cheap LLM (October 2026) · Best open-weights LLM (October 2026) · Fastest LLM (October 2026)
Quality against…
Showing 24 models: the frontier no other model beats on both axes, the top eight by index, the best model in each price band, and anything new in the last seven days. Up and to the left is better: more quality for less money. Hover for details, click to open a model.
Claude Opus 5.5max
Gemini 4 ArgonhighClaude Fable 5.1max
GPT-6 AstramaxGPT-6.1 SolmaxClaude Sonnet 5.5maxClaude Fable 5maxClaude Opus 5max
Muse Spark 1.3max
Kimi K3.1 ProGemini 3.8 Flashhigh
DeepSeek V4.1 Flash 1017max
MiMo-V2.6-Pro
GLM 5.3 Flash
MiniMax M3.1 FlashClaude Haiku 5.5xhigh
Step 5 PreviewMiMo-V2.6-FlashMiMo-V2.5Step 3.5 Flash
Ling 3.0 Flash
Qwen3.5 FlashQwen3.5-4Bthinking
Granite 4.2 3B- Claude Opus 5.5 (max)
- Gemini 4 Argon (high)
- Claude Fable 5.1 (max)
- GPT-6 Astra (max)
- GPT-6.1 Sol (max)
- Claude Sonnet 5.5 (max)
- Claude Fable 5 (max)
- Claude Opus 5 (max)
- Muse Spark 1.3 (max)
- Kimi K3.1 Pro
- Gemini 3.8 Flash (high)
- DeepSeek V4.1 Flash 1017 (max)
- MiMo-V2.6-Pro
- GLM 5.3 Flash
- MiniMax M3.1 Flash
- Claude Haiku 5.5 (xhigh)
- Step 5 Preview
- MiMo-V2.6-Flash
- MiMo-V2.5
- Step 3.5 Flash
- Ling 3.0 Flash
- Qwen3.5 Flash
- Qwen3.5-4B (thinking)
- Granite 4.2 3B
Straight answers
The questions people arrive with, answered from today’s data. Top five each; the board behind it is one click away.
- Best under $2/M
Highest quality with a blended price below $2 per million tokens.
- 1Gemini 3.8 Flashhigh$1.50/M
- 2MiMo-V2.6-Pro$0.55/M
- 3DeepSeek V4.1 Flash 1017max$0.53/M
- 4Gemini 3.7 Flashhigh$1.50/M
- 5GLM 5.3 Flash$0.24/M
- 1
- Best under $0.50/M
Highest quality for budget workloads, below fifty cents per million tokens.
- 1GLM 5.3 Flash$0.238/M
- 2MiniMax M3.1 Flash$0.262/M
- 3Claude Haiku 5.5xhigh$0.200/M
- 4Qwen3.8-Flash-Next$0.230/M
- 5GPT-6 Lunamax$0.200/M
- 1
- Fast and good
At least 100 tokens per second with an index above 55.
- 1
K2 Horizon 375B A23B909 tok/s - 2
Grok 4.20 Multi-Agent379 tok/s - 3Muse Spark 1.2xhigh307 tok/s
- 4GPT-5.4 Minixhigh255 tok/s
- 5Step 5 Preview229 tok/s
- 1
- Quick answers
Index above 55 and the answer starts within two seconds.
- 1Gemma 4 26B A4B0.5 s
- 2Claude Sonnet 4.60.7 s
- 3Qwen3.5 Plusthinking1.0 s
- 4Claude Sonnet 4.5high1.1 s
- 5Claude Opus 4.6high1.2 s
- 1
- Best open weights
Highest-quality models you can download and run yourself.
- 1Kimi K3max64.8
- 2MiMo-V2.6-Pro64.0
- 3DeepSeek V4.1 Flash 1017max64.0
- 4GLM 5.3max63.5
- 5GLM 5.3 Flash62.7
- 1
- Cheapest 1M context
Context window of at least one million tokens, ordered by price.
- 1Qwen3.5 Flash$0.09/M
- 2
Nemotron 3.5 Lightning$0.10/M - 3DeepSeek V4 Flash 0420max$0.14/M
- 4
Agnes 2.5 Pro Beta$0.15/M - 5Qwen-Plus$0.16/M
- 1
- Best for coding
Highest coding category score.
- 1Claude Opus 5.5max81.9
- 2Claude Fable 5.1max74.9
- 3GPT-6 Astramax73.9
- 4Claude Fable 5max73.8
- 5Muse Wonder Fullmax72.1
- 1
- Best for agents
Highest agents-and-tools category score, among models with at least three agentic results.
- 1Claude Sonnet 5.5max70.1
- 2Kimi K3.1 Pro68.0
- 3GPT-5.6 Solmax66.5
- 4Muse Spark 1.3max65.6
- 5Claude Opus 4.765.4
- 1
Price cuts
Listed blended prices that fell by more than 5% in the last 90 days.
| Model | Before | After | Change | When |
|---|---|---|---|---|
| Qwen3.5 Plus | $0.900/M | $0.258/M | −71% | 25 Oct 2026 |
| GLM 5V Turbo | $1.90/M | $1.38/M | −27% | 25 Oct 2026 |
| Qwen3.5 Flash | $0.175/M | $0.093/M | −47% | 25 Oct 2026 |
| Qwen3 32B | $1.22/M | $0.502/M | −59% | 25 Oct 2026 |
| GLM 4.5V | $0.900/M | $0.448/M | −50% | 25 Oct 2026 |
| Qwen3 235B A22B | $1.22/M | $0.502/M | −59% | 25 Oct 2026 |
| Qwen-Plus | $0.600/M | $0.158/M | −74% | 25 Oct 2026 |
| Qwen3 14B | $0.612/M | $0.251/M | −59% | 25 Oct 2026 |
| Qwen3.6 Flash | $0.563/M | $0.371/M | −34% | 25 Oct 2026 |
| Qwen3 8B | $0.310/M | $0.126/M | −59% | 25 Oct 2026 |
| Qwen Turbo | $0.088/M | $0.055/M | −37% | 25 Oct 2026 |
| Qwen3.7 Flash | $0.055/M | $0.052/M | −6% | 25 Oct 2026 |
| Qwen-VL | $1.40/M | $0.316/M | −77% | 25 Oct 2026 |
| Qwen-VL Plus | $0.315/M | $0.158/M | −50% | 25 Oct 2026 |
| Qwen3.8 Flash | $0.230/M | $0.189/M | −18% | 25 Oct 2026 |
| QwQ Plus | $1.20/M | $0.316/M | −74% | 25 Oct 2026 |
| GLM 5.3 FlashX | $0.590/M | $0.486/M | −18% | 25 Oct 2026 |
| Qwen Flash | $0.138/M | $0.071/M | −49% | 25 Oct 2026 |
| Qwen-MT Plus | $3.69/M | $0.388/M | −89% | 25 Oct 2026 |
| Qwen-MT Turbo | $0.242/M | $0.146/M | −40% | 25 Oct 2026 |
Core boards
Six ways to rank the same models. Each opens as a chart plus a searchable, sortable table.
- Quality: the BenchLeader Index
Composite of every quality benchmark, normalised per benchmark and averaged per category. Requires results on at least five benchmarks across two categories.
- Value: quality per dollar
BenchLeader Index divided by the blended price per million tokens (3 input : 1 output). Higher means more quality for the money.
- Speed: output tokens per second
Output tokens per second: Artificial Analysis' measured median where it covers a model, otherwise OpenRouter's live-traffic median smoothed over the last seven days.
- Latency: time to first answer
Median seconds until the answer starts, including any reasoning phase (Artificial Analysis), or time to first streamed token where that is all we have.
- Cost: blended price per million tokens
First-party list price where published, blended at 3 input tokens per output token.
- Token efficiency: quality per dollar of real work
BenchLeader Index divided by what it cost Artificial Analysis to run its whole Intelligence Index on that configuration. Unlike price per token this counts the tokens a model actually burns, so a model that thinks at length scores lower than its sticker price suggests.
- Cost to run a full benchmark suite
What Artificial Analysis actually paid to run its whole Intelligence Index on each configuration: a fixed batch of work, billed at list price. The spread between models is far wider than their per-token prices, because verbose reasoning multiplies the token count.
- Response time: seconds to a complete answer
Median seconds from request to the last token, reasoning included (Artificial Analysis). Time to first answer tells you when output starts; this tells you when it ends.
- Context window
Maximum context length in tokens as listed by the provider.
Per task
Prices per million tokens hide the shape of real work. These boards price and time five representative workloads from each model’s list price, output speed and time to first token.
| Workload | Input tokens | Output tokens | Boards |
|---|---|---|---|
Chat reply A short assistant turn: 400 tokens of context in, 300 tokens out. | 400 | 300 | CostTime |
Summarize a 30-page report 12,000 tokens of document in, a 600-token summary out. | 12,000 | 600 | CostTime |
Code edit 6,000 tokens of files and instructions in, a 1,500-token diff out. | 6,000 | 1,500 | CostTime |
Agentic coding session One long tool-using loop: 60,000 tokens in across turns, 4,000 tokens out. | 60,000 | 4,000 | CostTime |
Structured extraction 2,000 tokens of source text in, 200 tokens of JSON out. | 2,000 | 200 | CostTime |
Questions people ask
- Which LLM leaderboard should I use?
- Start from the question you are asking. For overall quality use the quality board (the BenchLeader Index); for a budget, the cost or value boards; for chat and voice, speed and latency; for long documents, context. The per-task boards price five typical workloads.
- How often are the leaderboards updated?
- Every board is rebuilt once a day (the run starts at 22:03 UTC); the current data is from 25 Oct 2026.
- What is the fastest model on the leaderboard, and what does it cost?
- On 25 Oct 2026, Celeris-1 is the fastest at 1581 tokens per second, for $0.325 per 1M tokens blended, with a BenchLeader Index of 41.4.