Top for agents
LMSpeed is an AI model directory for comparing API price, output speed, first-token latency, provider coverage, and benchmark data. Use it to narrow your model and provider choice, then test the endpoint that fits your workload.
Price, speed, latency, and availability can change. Treat this table as a current signal and verify your own endpoint before you deploy.
| Release date | |||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Context262.1K | Input— | Output— | Providers |
Agents, Coding, Reasoning, and the other capability columns are 0–100 observed-capability estimates relative to the eligible model population in a dated Category Score V3 run.
They are not success rates, IQ scores, or an average across all eight categories. Read them with the 80% interval, measured dimensions, benchmark families, and evidence shown on each model page.
Each row brings together the information you need to compare an LLM API. Some fields are blank when LMSpeed has no current data for that model or provider.
Start with the decision that matters most for your workload, then compare the current rows before you test an endpoint.
LMSpeed runs standardized five-round API speed tests on each model, measuring output throughput (tokens per second), first-token latency, and total response time across multiple providers.
Latency varies by provider and model. Use the LMSpeed model directory to sort by latency and find the model with the fastest first-token response time. Check the latency leaderboard for monthly rankings.
LMSpeed lists input and output token prices per million for each model across available providers. Sort by price to find a lower-cost option, or filter by provider to compare rates side by side.
— |
— |
— |
— |
— |
— |
— |
— |
| Throughput — | Latency — |
| Release date2025-12-23 |
|
| Seed 1.6 FlashByteDance Seed | Context262.1K | Input— | Output— | Providers | — | — | — | — | — | — | — | — | Throughput — | Latency — | Release date2025-12-23 |
|
| GPT-5.2 CodexOpenAI | Context400K | Input$1.75/M | Output$14.00/M | Providers +151 | 51.7±9.3E | 58.1±11.8E | 59.8±10.8E | 52.1±16.0P | — | — | — | 55.8±16.0P | Throughput 54 t/s | Latency 3.51s | Release date2026-01-14 |
|
| GLM-4.7 FlashZ.ai | Context200K | Input$0.070/M | Output$0.400/M | Providers +44 | — | 42.2±16.0P | 40.9±13.9P | — | — | — | — | — | Throughput 49 t/s | Latency 23.19s | Release date2026-01-19 |
|
| LFM 2.5 1.2B InstructLiquidAI | Context32.8K | Input— | Output— | Providers +15 | — | — | — | — | — | — | — | — | Throughput — | Latency — | Release date2026-01-20 |
|
| Solar Pro 3Upstage | Context128K | Input— | Output— | Providers +1 | — | 41.5±16.0P | 49.2±13.9P | — | — | — | — | — | Throughput 41 t/s | Latency 10.64s | Release date2026-01-27 |
|
| Kimi K2.5MoonshotAI | Context262.1K | Input$0.600/M | Output$3.00/M | Providers +183 | 40.3±5.1 | 49±7.8 | 56.1±8.3 | 47.1±16.3P | 49.2±9.1E | 44.4±12.2E | 58.1±12.0E | 54.2±11.9E | Throughput 144 t/s | Latency 11.93s | Release date2026-01-27 |
|
| Step 3.5 FlashStepFun | Context262.1K | Input$0.100/M | Output$0.300/M | Providers +61 | — | 53.3±16.0P | 54.9±13.9P | — | — | — | — | — | Throughput 104 t/s | Latency 6.74s | Release date2026-01-29 |
|
| Qwen3 Coder NextQwen | Context262.1K | Input$0.350/M | Output$1.20/M | Providers +12 | — | 47.6±16.0P | 49.2±13.9P | — | — | — | — | — | Throughput — | Latency — | Release date2026-02-04 |
|
| Claude Opus 4.6Anthropic | Context1M | Input$5.00/M | Output$25.00/M | Providers +213 | 55.7±5.7 | 59.6±8.0 | 54.1±8.7 | 53.7±13.7P | 56.5±16.1P | 54±17.3P | 49±11.2E | 44.8±16.0P | Throughput 46 t/s | Latency 5.36s | Release date2026-02-04 |
|
| Qwen3 Max ThinkingQwen | Context262.1K | Input$1.20/M | Output$6.00/M | Providers +3 | — | 52.6±13.9P | 55.6±11.1E | — | 51.7±16.1P | — | — | — | Throughput — | Latency — | Release date2026-02-09 |
|
| GLM-5Z.ai | Context202.8K | Input$1.00/M | Output$3.20/M | Providers +172 | 42.8±5.6 | 52.3±8.1 | 49.5±8.1 | 43.3±16.3P | 51.1±9.1E | 44±12.2E | — | 52.6±11.9E | Throughput 46 t/s | Latency 22.00s | Release date2026-02-11 |
|
| MiniMax M2.5MiniMax | Context204.8K | Input$0.300/M | Output$1.20/M | Providers +175 | — | 50.4±11.8E | 55.5±13.9P | — | — | — | — | — | Throughput 57 t/s | Latency 9.85s | Release date2026-02-12 |
|
| Qwen3.5 397B A17BQwen | Context262.1K | Input$0.600/M | Output$3.60/M | Providers | — | 54.4±16.0P | 59±13.9P | — | — | — | — | — | Throughput — | Latency — | Release date2026-02-16 |
|
| Claude Sonnet 4.6Anthropic | Context1M | Input$3.00/M | Output$15.00/M | Providers +207 | 53.6±6.1 | 56.5±8.6 | 50.3±8.7 | 70.6±16.3P | 53.2±16.1P | — | 44.5±16.2P | 43.8±16.0P | Throughput 43 t/s | Latency 4.13s | Release date2026-02-17 |
|
| Gemini 3.1 ProGoogle | Context1.0M | Input— | Output— | Providers +172 | 47.9±5.8 | 56±8.7 | 59.8±8.8 | 57.8±15.1P | 55.3±16.1P | 59.7±17.3P | 56±6.7 | 55.6±16.0P | Throughput 92 t/s | Latency 15.36s | Release date2026-02-19 |
|
| Gemini 3.1 Pro PreviewGoogle | Context1.0M | Input$2.00/M | Output$12.00/M | Providers +8 | — | 65.5±16.0P | 64.9±13.9P | — | — | — | — | — | Throughput — | Latency — | Release date2026-02-19 |
|
| Aion 2.0AionLabs | Context131.1K | Input— | Output— | Providers | — | — | — | — | — | — | — | — | Throughput — | Latency — | Release date2026-02-23 |
|
| GPT-5.3 CodexOpenAI | Context400K | Input$1.75/M | Output$14.00/M | Providers +242 | 53.5±6.6 | 59.2±8.5 | 60.7±10.8E | 55.3±16.0P | — | — | — | 54.9±16.0P | Throughput 78 t/s | Latency 3.91s | Release date2026-02-24 |
|
| Qwen3.5 Flash | Context1M | Input— | Output— | Providers +67 | — | — | — | — | 42.7±16.1P | — | — | — | Throughput 115 t/s | Latency 7.92s | Release date2026-02-25 |
|
| Qwen3.5-122B-A10BQwen | Context262.1K | Input$0.400/M | Output$3.20/M | Providers | 49.7±10.6E | 49.4±11.8E | 53.6±8.3 | 43.7±16.3P | — | 45.4±16.4P | 47.8±9.4E | 54.5±11.9E | Throughput — | Latency — | Release date2026-02-25 |
|
| Qwen3.5-27BQwen | Context262.1K | Input$0.300/M | Output$2.40/M | Providers | 47.1±8.7 | 49.7±11.2E | 54.1±8.3 | 41.4±16.3P | — | 45.4±16.4P | 49.9±12.1E | 57.3±11.9E | Throughput — | Latency — | Release date2026-02-25 |
|
| Qwen3.5-35B-A3BQwen | Context262.1K | Input$0.250/M | Output$2.00/M | Providers | 43.9±8.7 | 42.7±11.2E | 51.1±8.3 | 38±16.3P | — | 41.6±16.4P | 46±12.1E | 51.7±11.9E | Throughput — | Latency — | Release date2026-02-25 |
|
| Seed 2.0 MiniByteDance Seed | Context262.1K | Input— | Output— | Providers +1 | — | — | — | — | — | — | — | — | Throughput — | Latency — | Release date2026-02-26 |
|
| Gemini 3.1 Flash Lite PreviewGoogle | Context1.0M | Input$0.250/M | Output$1.50/M | Providers +3 | — | 54.3±16.0P | 53.9±13.9P | — | — | — | — | — | Throughput — | Latency — | Release date2026-03-03 |
|
| GPT-5.3OpenAI | Context128K | Input— | Output— | Providers +71 | — | — | — | — | — | — | — | — | Throughput 59 t/s | Latency 2.57s | Release date2026-03-03 |
|
| Mercury 2Inception | Context128K | Input$0.250/M | Output$0.750/M | Providers | — | 52.2±16.0P | 52.1±13.9P | — | — | — | — | — | Throughput 490 t/s | Latency 0.89s | Release date2026-03-04 |
|
| GPT-5.4OpenAI | Context1.1M | Input$2.50/M | Output$15.00/M | Providers +277 | 56.6±5.1 | 62.9±10.3E | 59.5±10.8E | 59.9±15.1P | 57.6±16.1P | — | 53.7±6.7 | 54.3±16.0P | Throughput 49 t/s | Latency 4.45s | Release date2026-03-05 |
|
| GPT-5.4 ProOpenAI | Context1.1M | Input$30.00/M | Output$180.00/M | Providers +65 | 60.5±16.1P | — | 62.2±11.4E | 62.2±16.1P | 56.8±16.1P | — | 73.1±16.1P | — | Throughput — | Latency — | Release date2026-03-05 |
|
| Qwen3.5 | Context262.1K | Input$0.030/M | Output$0.150/M | Providers +90 | 46±5.3 | 40.4±9.0 | 53.4±8.1 | 53.1±11.6E | 41.5±11.5E | 54.8±12.2E | 51.3±6.5 | 53.9±11.9E | Throughput 46 t/s | Latency 10.19s | Release date2026-03-10 |
|
| Qwen3.5-9BQwen | Context262.1K | Input$0.100/M | Output$0.150/M | Providers— | — | 43.9±16.0P | 52.5±13.9P | — | — | — | — | — | Throughput — | Latency — | Release date2026-03-10 |
|
| Seed 2.0 LiteByteDance Seed | Context262.1K | Input— | Output— | Providers +1 | — | — | — | — | — | — | — | — | Throughput — | Latency — | Release date2026-03-10 |
|
| Nemotron 3 SuperNVIDIA | Context1M | Input— | Output— | Providers +6 | — | — | — | — | — | — | — | — | Throughput — | Latency — | Release date2026-03-11 |
|
| GLM-5 TurboZ.ai | Context202.8K | Input— | Output— | Providers +70 | 52.5±11.9E | 55.5±16.0P | 52.2±10.8E | 48.8±16.0P | — | — | — | 54±16.0P | Throughput 56 t/s | Latency 15.05s | Release date2026-03-15 |
|
| GPT-5.4 MiniOpenAI | Context400K | Input$0.750/M | Output$4.50/M | Providers +225 | 51.5±8.1 | 57.7±11.8E | 56.7±10.8E | 48.1±14.0P | 49.7±16.1P | — | 46.3±16.1P | 54.1±16.0P | Throughput 134 t/s | Latency 3.54s | Release date2026-03-17 |
|
| GPT-5.4 NanoOpenAI | Context400K | Input$0.200/M | Output$1.25/M | Providers +88 | 45.5±8.1 | 50±11.8E | 50.2±10.8E | 44.4±14.0P | 51.8±16.1P | — | 28.1±16.1P | 55.1±16.0P | Throughput 137 t/s | Latency 3.38s | Release date2026-03-17 |
|
| MiniMax M2.7MiniMax | Context204.8K | Input$0.300/M | Output$1.20/M | Providers +159 | 45.6±6.0 | 47.7±8.2 | 54.8±10.8E | 52.9±16.0P | — | — | — | 55±16.0P | Throughput 178 t/s | Latency 7.32s | Release date2026-03-18 |
|
| Kat Coder Pro V2Kwaipilot | Context256K | Input$0.300/M | Output$1.20/M | Providers +4 | — | 51.9±16.0P | 55±13.9P | — | — | — | — | — | Throughput — | Latency — | Release date2026-03-27 |
|
| Grok 4.20xAI | Context2M | Input$2.00/M | Output$6.00/M | Providers +85 | 43.6±11.2E | 47.8±8.6 | 52.9±11.3E | 44.1±15.1P | — | — | 38.3±9.0E | — | Throughput 75 t/s | Latency 4.82s | Release date2026-03-31 |
|
| Grok 4.20 Multi AgentxAI | Context2M | Input— | Output— | Providers +28 | — | — | — | — | — | — | — | — | Throughput — | Latency — | Release date2026-03-31 |
|
| Trinity Large ThinkingArcee AI | Context262.1K | Input$0.235/M | Output$0.875/M | Providers +1 | 44.5±9.3E | 50.4±16.0P | 44.5±10.8E | 41.3±16.0P | — | — | — | 48.4±16.0P | Throughput 108 t/s | Latency 1.94s | Release date2026-04-01 |
|
| GLM-5V TurboZ.ai | Context202.8K | Input— | Output— | Providers +33 | 47.4±9.3E | 55.4±16.0P | 51.1±10.8E | 47.8±16.0P | — | — | — | 49.9±16.0P | Throughput — | Latency — | Release date2026-04-01 |
|
| Qwen3.6 PlusQwen | Context1M | Input$0.500/M | Output$3.00/M | Providers +116 | 50.5±5.5 | 53±8.2 | 58±8.1 | 52±11.6E | 52.2±9.1E | 52.4±12.2E | 51.2±6.5 | 55.9±11.9E | Throughput 47 t/s | Latency 24.53s | Release date2026-04-02 |
|
| GLM-5.1Z.ai | Context202.8K | Input$1.40/M | Output$4.40/M | Providers +173 | 54.3±5.7 | 55±8.8 | 54.3±10.8E | 57.1±14.0P | 51.7±9.1E | — | — | 55.3±16.0P | Throughput 48 t/s | Latency 14.18s | Release date2026-04-07 |
|
| Kimi K2.6MoonshotAI | Context262.1K | Input$0.950/M | Output$4.00/M | Providers +114 | 53.6±5.1 | 56.8±6.1 | 55.5±10.8E | 50.6±14.0P | 59.1±9.1E | — | 54.3±9.0E | 55.2±16.0P | Throughput 53 t/s | Latency 18.41s | Release date2026-04-20 |
|
| Ling 2.6 FlashinclusionAI | Context262.1K | Input$0.100/M | Output$0.300/M | Providers +2 | 45.4±11.8E | 43.5±16.0P | 39.3±10.8E | 35.4±14.0P | — | — | — | 48.7±16.0P | Throughput — | Latency — | Release date2026-04-21 |
|
| MiMo-V2.5Xiaomi | Context1.0M | Input$0.140/M | Output$0.280/M | Providers +94 | 50.9±8.9 | 51.2±9.3E | 56.8±13.9P | — | — | — | 49.3±11.9E | — | Throughput 81 t/s | Latency 3.64s | Release date2026-04-22 |
|
| MiMo-V2.5-ProXiaomi | Context1.0M | Input$0.435/M | Output$0.870/M | Providers +98 | 52.9±6.0 | 49.6±9.3E | 54.9±10.8E | 55.2±14.0P | — | — | — | 56.9±16.0P | Throughput 45 t/s | Latency 5.86s | Release date2026-04-22 |
|
| Hy3 previewTencent | Context262.1K | Input$0.123/M | Output$0.430/M | Providers +2 | 47±11.2E | 48.3±9.3E | 52.1±10.8E | 41.4±14.0P | — | — | — | 50.5±16.0P | Throughput — | Latency — | Release date2026-04-22 |
|
| Ling 2.6 1TinclusionAI | Context262.1K | Input$0.300/M | Output$2.50/M | Providers | — | 51±16.0P | 49.1±13.9P | — | — | — | — | — | Throughput — | Latency — | Release date2026-04-23 |
|
| DeepSeek V4 FlashDeepSeek | Context1.0M | Input$0.140/M | Output$0.280/M | Providers +190 | 48.5±5.3 | 44.5±6.6 | 51.6±8.4 | 39.1±14.0P | 36.6±12.4E | — | — | 54.2±16.0P | Throughput 66 t/s | Latency 7.28s | Release date2026-04-24 |
|
| DeepSeek V4 ProDeepSeek | Context1.0M | Input$0.435/M | Output$0.870/M | Providers +182 | 52.6±5.2 | 49.4±6.3 | 53.8±8.4 | 44±14.0P | 35.9±12.4E | — | — | 53.3±16.0P | Throughput 37 t/s | Latency 8.37s | Release date2026-04-24 |
|
| GPT-5.5OpenAI | Context1.1M | Input$5.00/M | Output$30.00/M | Providers +110 | 62.5±5.0 | 60±8.7 | 61.4±8.8 | 60.6±14.0P | 56.9±16.1P | — | 56.4±16.1P | 55.1±16.0P | Throughput 44 t/s | Latency 5.99s | Release date2026-04-24 |
|
| GPT-5.5 ProOpenAI | Context1.1M | Input— | Output— | Providers +24 | 61.3±16.1P | — | 64.5±14.0P | 61.1±16.1P | 57.2±16.1P | — | — | — | Throughput — | Latency — | Release date2026-04-24 |
|
| Qwen3.6 27BQwen | Context262.1K | Input$0.600/M | Output$3.60/M | Providers +3 | 56.5±6.6 | 46.4±6.0 | 54.8±8.7 | 37.3±11.6E | 41.3±11.5E | — | 46.4±6.5 | 52±16.0P | Throughput — | Latency — | Release date2026-04-27 |
|
| Qwen3.6 Max PreviewQwen | Context262.1K | Input$1.30/M | Output$7.80/M | Providers +14 | 54.5±11.8E | 53.2±8.9 | 56.9±10.8E | 55.1±16.3P | 50.6±16.1P | — | — | 55.4±16.0P | Throughput — | Latency — | Release date2026-04-27 |
|
| Qwen3.6 35B A3BQwen | Context262.1K | Input$0.248/M | Output$1.49/M | Providers +5 | 47.3±5.9 | 38.9±6.0 | 53±8.7 | 39.1±12.2E | 34.7±11.5E | — | 42.8±7.1 | 50.9±16.0P | Throughput — | Latency — | Release date2026-04-27 |
|
| Qwen3.5 PlusAlibaba | Context1M | Input— | Output— | Providers +106 | 44.8±16.2P | 46.5±16.1P | — | — | 48.8±16.1P | — | — | — | Throughput 51 t/s | Latency 13.81s | Release date2026-04-27 |
|
| Gemini FlashGoogle | Context1.0M | Input— | Output— | Providers +41 | — | — | — | — | — | — | — | — | Throughput — | Latency — | Release date2026-04-27 |
|
| Gemini ProGoogle | Context1.0M | Input— | Output— | Providers +36 | — | — | — | — | — | — | — | — | Throughput — | Latency — | Release date2026-04-27 |
|
An AI model directory is a searchable list of models and their comparison data. On LMSpeed, it brings together API price, throughput, first-token latency, provider coverage, and capability data when available.
Start with the metric that matters most for your workload. Then open a model page to compare provider data. Run your own speed test before you use an endpoint in production.
Yes. Price, availability, throughput, and latency can change by model, provider, and time. Use current page values as a comparison signal and confirm with your own endpoint test.
