Choose a model to open its comparison page.
The page also shows measured API speed and first-token latency.
DeepSeek Chat V3.1 is an instruction-tuned variant in the DeepSeek series, optimized for following instructions and conversational tasks.
Input and output token limits for this model, plus how it ranks on long-context understanding.
Third-party OpenRouter endpoint data, shown separately from LMSpeed measurements. Some 30-minute live performance fields only appear after syncing with an OpenRouter API key.
| Provider endpoint | Input | Output | 1d uptime | 30m latency | 30m throughput | Context / output |
|---|---|---|---|---|---|---|
Novita novita/fp8 | $0.270/M | $1/M | 100.0% | — | — | 131.1K tokens / 32.8K tokens |
AtlasCloud atlas-cloud/fp8 | $0.300/M | $0.950/M | 99.8% | — | — | 131.1K tokens / 65.5K tokens |
SambaNova sambanova/fp8 | $0.650/M | $1.50/M | 99.8% | — | — | 131.1K tokens / 7.2K tokens |
DeepInfra deepinfra/fp4 | $0.210/M | $0.790/M | 99.7% | — | — | 163.8K tokens / 32.8K tokens |
Google google-vertex/us-west2 | $0.600/M | $1.70/M | 99.6% | — | — | 163.8K tokens / 32.8K tokens |
WandB wandb/fp8 | $0.550/M | $1.65/M | 98.6% | — | — | 161K tokens / 161K tokens |
SiliconFlow siliconflow/fp8 | $0.270/M | $1/M | 97.2% | — | — | 163.8K tokens / 163.8K tokens |
Mara mara | $0.600/M | $1.70/M | 88.1% | — | — | 131.1K tokens / 7.2K tokens |
gemini-2-5-pro
Google Gemini 2.5 Pro is Google advanced multimodal model with a 1M-token context window, strong STEM reasoning, and native support for images, audio, and video understanding.
qwen3
Alibaba Qwen3 is the Qwen family's flagship LLM series with dense and MoE variants, seamless thinking/non-thinking modes, and leading open-source performance in math, code, and agent tasks.
kimi-k2
Moonshot Kimi K2 is a trillion-parameter MoE model from Moonshot AI, built for long-context chat, agentic tool use, and high-quality Chinese-English bilingual generation.
qwen3-coder
Alibaba Qwen3 Coder is a code-specialized variant in the Qwen series, optimized for code generation, debugging, and software development tasks.
gpt-5-4
OpenAI GPT-5.4 extends the GPT-5 family with stronger instruction following, deeper tool use, and improved performance on coding, math, and long-document analysis.
gpt-5-3-codex
OpenAI GPT-5.3 Codex is a code-specialized variant in the GPT-5 series, optimized for code generation, debugging, and software development tasks.
