Models
Every model routable through LokaRouter, with pricing, throughput, and latency at a glance.
| Model | Modality | Context | Input / MTok | Output / MTok | Throughput | Latency | Providers |
|---|---|---|---|---|---|---|---|
openai/gpt-5.2Current flagship for reasoning, coding, and agentic work. | TextImage | 400K | $1.75 | $14.00 | 118 tok/s | 620 ms | 6 |
openai/gpt-5.2-proExtended-compute GPT-5.2 for the hardest problems. | TextImage | 400K | $15.00 | $120.00 | 76 tok/s | 940 ms | 4 |
anthropic/claude-opus-4.6Deepest reasoning Claude with adaptive thinking budgets. | TextImage | 200K | $5.00 | $25.00 | 92 tok/s | 760 ms | 4 |
anthropic/claude-sonnet-4.6Balanced production flagship with stronger code gains. | TextImage | 200K | $3.00 | $15.00 | 104 tok/s | 620 ms | 5 |
google/gemini-3-proCurrent Gemini flagship with deep-think mode. | TextImageAudioVideo | 1M | $2.00 | $12.00 | 112 tok/s | 590 ms | 5 |
openai/gpt-5.1Previous flagship with adaptive reasoning and warm tuning. | TextImage | 400K | $1.25 | $10.00 | 116 tok/s | 600 ms | 6 |
google/gemini-3-flashSpeed flagship: Pro-class quality at Flash latency. | TextImageAudioVideo | 1M | $0.50 | $3.00 | 178 tok/s | 330 ms | 6 |
openai/gpt-5.1-codex-maxLong-horizon agentic coding flagship for real repos. | TextImage | 400K | $1.25 | $10.00 | 112 tok/s | 640 ms | 5 |
anthropic/claude-sonnet-4.5Proven all-round Sonnet for production workloads. | TextImage | 200K | $3.00 | $15.00 | 102 tok/s | 640 ms | 5 |
openai/gpt-5-miniCost-efficient GPT-5 tier for high-volume tasks. | TextImage | 400K | $0.25 | $2.00 | 165 tok/s | 480 ms | 5 |
google/gemini-2.5-flashAdaptive-thinking multimodal flash tier. | TextImageAudio | 1M | $0.30 | $2.50 | 178 tok/s | 330 ms | 6 |
anthropic/claude-opus-4.5Previous Opus flagship, still favored for long agent runs. | TextImage | 200K | $5.00 | $25.00 | 88 tok/s | 780 ms | 4 |
meta/llama-4.1-maverick-405bOpen MoE flagship of the Llama 4.1 family. | TextImage | 1M | $0.28 | $0.95 | 148 tok/s | 380 ms | 7 |
anthropic/claude-haiku-4.5Fast, inexpensive near-flagship quality. | TextImage | 200K | $1.00 | $5.00 | 158 tok/s | 410 ms | 5 |
deepseek/deepseek-v4Sparse-attention flagship for coding and agents. | Text | 1M | $0.35 | $1.40 | 132 tok/s | 420 ms | 6 |
google/gemini-2.5-proMillion-token thinking model in wide production use. | TextImageAudio | 1M | $1.25 | $10.00 | 112 tok/s | 590 ms | 5 |
deepseek/deepseek-r2Second-generation open reasoning model trained with RL. | Text | 1M | $0.55 | $2.20 | 104 tok/s | 780 ms | 5 |
meta/llama-4.1-scout-109bEfficient multimodal MoE for broad deployment. | TextImage | 1M | $0.18 | $0.65 | 163 tok/s | 350 ms | 8 |
qwen/qwen3.5-plusFlagship 1M-context multimodal tier for agentic work. | TextImage | 1M | $0.90 | $4.50 | 124 tok/s | 440 ms | 4 |
openai/gpt-5.2-chatNon-reasoning tuning of GPT-5.2 for conversation and drafting. | TextImage | 400K | $1.75 | $14.00 | 128 tok/s | 540 ms | 4 |
meta/llama-4.1-behemoth-2tTrillion-parameter MoE teacher model now served broadly. | TextImage | 1M | $0.75 | $1.00 | 98 tok/s | 520 ms | 5 |
deepseek/deepseek-v3.2Cost flagship with sparse attention and wide adoption. | Text | 164K | $0.27 | $1.10 | 132 tok/s | 420 ms | 6 |
qwen/qwen3-coder-nextOpen MoE code model with 256K context for agentic coding. | Text | 262K | $0.14 | $0.50 | 150 tok/s | 360 ms | 5 |
qwen/qwen3.5-flashFast tier with strong instruction following. | TextImage | 1M | $0.15 | $0.90 | 176 tok/s | 320 ms | 5 |
openai/gpt-5.1-codex-miniCompact codex tier for fast, cheap code edits. | Text | 400K | $0.25 | $2.00 | 172 tok/s | 440 ms | 5 |
google/gemini-2.5-flash-liteCheapest Gemini tier for high-volume routing. | TextImage | 1M | $0.10 | $0.40 | 205 tok/s | 290 ms | 5 |
openai/gpt-4.1Million-token workhorse still in wide production use. | TextImage | 1M | $2.00 | $8.00 | 105 tok/s | 560 ms | 5 |
qwen/qwen3.5-397b-a17bOpen MoE flagship of the Qwen3.5 lineup. | TextImage | 262K | $0.40 | $1.40 | 138 tok/s | 380 ms | 6 |
meta/llama-4-scoutWidely deployed multimodal Llama 4. | TextImage | 328K | $0.15 | $0.60 | 163 tok/s | 350 ms | 8 |
x-ai/grok-4.2Latest Grok flagship with real-time knowledge hooks. | TextImage | 256K | $3.00 | $15.00 | 94 tok/s | 760 ms | 3 |
qwen/qwen3-maxLargest Qwen3 tier for complex agentic work. | Text | 262K | $1.20 | $6.00 | 118 tok/s | 470 ms | 4 |
openai/gpt-5-nanoSmallest GPT-5 tier for latency-sensitive workloads. | Text | 400K | $0.050 | $0.40 | 210 tok/s | 350 ms | 4 |
mistralai/mistral-large-3European open-weights flagship for reasoning and code. | Text | 256K | $2.00 | $6.00 | 110 tok/s | 520 ms | 5 |
google/gemini-2.5-flash-imageConversational image generation and editing. | TextImage | 33K | $0.30 | $2.50 | 96 tok/s | 640 ms | 4 |
openai/gpt-4.1-miniAffordable long-context mini in current deployments. | TextImage | 1M | $0.40 | $1.60 | 152 tok/s | 430 ms | 5 |
qwen/qwen3-coder-480bOpen code flagship for agentic coding. | Text | 262K | $0.29 | $1.20 | 129 tok/s | 400 ms | 6 |
mistralai/codestral-25.08Latest low-latency code completion specialist. | Text | 256K | $0.30 | $0.90 | 187 tok/s | 240 ms | 5 |
openai/gpt-5.1-chatConversational 5.1 tuning for support and companions. | TextImage | 400K | $1.25 | $10.00 | 132 tok/s | 520 ms | 4 |
x-ai/grok-4.1Frontier Grok generation with strong agentic tool use. | TextImage | 256K | $3.00 | $15.00 | 96 tok/s | 740 ms | 3 |
mistralai/mistral-medium-3.1Frontier-class quality at mid-tier pricing. | Text | 128K | $0.40 | $2.00 | 139 tok/s | 420 ms | 5 |
deepseek/deepseek-v3.1-terminusStabilized V3.1 with better agent reliability. | Text | 164K | $0.55 | $1.68 | 118 tok/s | 470 ms | 5 |
google/gemini-3-deep-researchAutonomous research agent producing cited reports. | TextImage | 1M | $2.00 | $12.00 | 46 tok/s | 1900 ms | 3 |
qwen/qwen3-coder-plusProprietary code model with repo-scale context. | Text | 1M | $0.45 | $1.80 | 124 tok/s | 420 ms | 4 |
x-ai/grok-4.1-fastSpeed tier with 2M context for agents. | TextImage | 2M | $0.20 | $0.50 | 172 tok/s | 290 ms | 3 |
qwen/qwen3-vl-235bOpen multimodal flagship for document agents. | TextImageVideo | 262K | $0.30 | $1.20 | 122 tok/s | 430 ms | 4 |
x-ai/grok-code-fast-1Cheap, fast coding model for agentic loops. | Text | 256K | $0.15 | $0.60 | 190 tok/s | 280 ms | 3 |
anthropic/claude-3.7-sonnetHybrid-reasoning Sonnet still common in enterprises. | TextImage | 200K | $3.00 | $15.00 | 96 tok/s | 660 ms | 4 |
openai/gpt-image-1.5Latest instruction-following image generation. | Image | 32K | $5.00 | $40.00 | 3 tok/s | 3800 ms | 3 |
mistralai/magistral-2Transparent reasoning model with visible planning. | Text | 128K | $2.00 | $5.00 | 104 tok/s | 690 ms | 3 |
google/gemini-2.0-flashStable previous-gen flash still common in pipelines. | TextImageAudio | 1M | $0.10 | $0.40 | 196 tok/s | 310 ms | 4 |
Sample data for demonstration — connect your workspace for live stats.