Models

x-aiLLM models1M context$2.50 / 1M input$5.00 / 1M outputtext, image input

Grok 4.3 is a reasoning model from xAI that accepts text and image inputs with text output, suited to agentic workflows, instruction-following, and applications requiring high factual accuracy. Reasoning effort is configurable (none, low, medium, or high), and a 1M-token context window with no output limit makes it well-suited to long-document analysis, deep research, and multi-step agentic tasks.

Published catalog burn rate

ANTHROPIC: Claude Opus 4.7

anthropicLLM models1M context$10.00 / 1M input$50.00 / 1M outputtext, image, file input

Claude Opus 4.7 is the next generation of Anthropic's Opus family, built for long-running, asynchronous agents. Building on Opus 4.6's coding and agentic strengths, it delivers stronger performance on complex, multi-step tasks such as large codebases, multi-stage debugging, and end-to-end project orchestration, plus improved knowledge work from document drafting to data analysis, maintaining coherence across very long outputs and extended sessions.

Published catalog burn rate

ANTHROPIC: Claude Sonnet 4.6

anthropicLLM models1M context$6.00 / 1M input$30.00 / 1M outputtext, image, file input

Claude Sonnet 4.6 is Anthropic's most capable Sonnet-class model yet, with frontier performance across coding, agents, and professional work. It excels at iterative development, complex codebase navigation, end-to-end project management with memory, polished document creation, and confident computer use for web QA and workflow automation.

Published catalog burn rate

OPENAI: GPT 5.5

openaiLLM models1.1M context$10.00 / 1M input$60.00 / 1M outputtext, image, file input

GPT-5.5 is OpenAI's frontier model for complex professional workloads, building on GPT-5.4 with stronger reasoning, higher reliability, and improved token efficiency on hard tasks. It pairs a 1M+ token context window with text and image inputs, enabling large-scale reasoning, coding, and multimodal workflows in a single system.

Published catalog burn rate

GOOGLE: Gemini 3 Flash

googleLLM models1M context$1.00 / 1M input$6.00 / 1M outputtext, image, audio, video, file input

Gemini 3 Flash is a high-speed, high-value thinking model designed for agentic workflows, multi-turn chat, and coding assistance. It delivers near-Pro reasoning and tool use at substantially lower latency, with a 1M-token context window and multimodal inputs including text, images, audio, video, and PDFs. It supports configurable thinking levels, structured output, tool use, and automatic context caching.

Published catalog burn rate

GOOGLE: Gemini 3.1 Pro

googleLLM models1M context$4.00 / 1M input$24.00 / 1M outputtext, image, audio, video, file input

Gemini 3.1 Pro is Google's frontier reasoning model, delivering enhanced software engineering performance, improved agentic reliability, and more efficient token usage. Building on the multimodal foundation of the Gemini 3 series, it combines high-precision reasoning across text, image, video, audio, and code with a 1M-token context window, and introduces a medium thinking level to balance cost, speed, and performance. It excels at agentic coding, structured planning, and multimodal analysis.

Published catalog burn rate

DEEPSEEK: DeepSeek V4 Pro

deepseekLLM models1M context$0.87 / 1M input$1.74 / 1M outputtext input

DeepSeek V4 Pro is a large-scale Mixture-of-Experts model with 1.6T total parameters (49B activated) and a 1M-token context window, built for advanced reasoning, coding, and long-horizon agent workflows. It uses a hybrid attention system for efficient long-context processing and supports high and xhigh reasoning efforts, suiting full-codebase analysis, multi-step automation, and large-scale synthesis.

Published catalog burn rate

DEEPSEEK: DeepSeek V4 Flash

deepseekLLM models1M context$0.20 / 1M input$0.39 / 1M outputtext input

DeepSeek V4 Flash is an efficiency-optimized Mixture-of-Experts model with 284B total parameters (13B activated) and a 1M-token context window, designed for fast inference and high-throughput workloads while maintaining strong reasoning and coding performance. Hybrid attention enables efficient long-context processing, with high and xhigh reasoning efforts, making it ideal for coding assistants, chat systems, and agent workflows.

Published catalog burn rate

MOONSHOTAI: Kimi K2.6

moonshotaiLLM models262K context$1.37 / 1M input$6.84 / 1M outputtext, image input

Kimi K2.6 is Moonshot AI's next-generation multimodal model, designed for long-horizon coding, coding-driven UI/UX generation, and multi-agent orchestration. It handles complex end-to-end coding across Python, Rust, and Go and turns prompts and visual inputs into production-ready interfaces, with an agent-swarm architecture that scales to hundreds of parallel sub-agents for autonomous task decomposition.

Published catalog burn rate

Z AI: GLM 5.1

z-aiLLM models203K context$1.96 / 1M input$6.16 / 1M outputtext input

GLM-5.1 delivers a major leap in coding capability, with especially strong gains on long-horizon tasks. Rather than minute-level interactions, it can work independently and continuously on a single task for more than eight hours, autonomously planning, executing, and refining its work to deliver complete, engineering-grade results.

Published catalog burn rate

Z AI: GLM 5.2

z-aiLLM models1M context$2.00 / 1M input$8.00 / 1M output

Published catalog burn rate

ANTHROPIC: Claude Sonnet 5

anthropicLLM models1M context$4.00 / 1M input$20.00 / 1M output

Published catalog burn rate

ANTHROPIC: Claude Fable 5

anthropicLLM models1M context$20.00 / 1M input$100.00 / 1M output

Published catalog burn rate

MOONSHOTAI: Kimi K3

moonshotaiLLM models1M context$6.00 / 1M input$30.00 / 1M output

Published catalog burn rate

GOOGLE: Gemini 3.5 Flash-Lite

googleLLM models1M context$0.60 / 1M input$5.00 / 1M output

Published catalog burn rate

GOOGLE: Gemini 3.6 Flash

googleLLM models1M context$3.00 / 1M input$15.00 / 1M output

Published catalog burn rate

ANTHROPIC: Claude Opus 5

anthropicLLM models1M context$10.00 / 1M input$50.00 / 1M output

Published catalog burn rate

ANTHROPIC: Claude Opus 5 (Fast)

anthropicLLM models1M context$20.00 / 1M input$100.00 / 1M output

Published catalog burn rate

QWEN: Qwen3.7 Flash

qwenLLM models1M context$0.06 / 1M input$0.26 / 1M output

Published catalog burn rate

QWEN: Qwen3.8 Max

qwenLLM models1M context$4.00 / 1M input$12.00 / 1M output

Published catalog burn rate

GOOGLE: Gemini 3.7 Flash

googleLLM models1M context$0.75 / 1M input$3.75 / 1M output

Published catalog burn rate

GOOGLE: Gemini 3.7 Flash (batch)

googleLLM models1M context$0.75 / 1M input$3.75 / 1M output

Published catalog burn rate

QWEN: Qwen3.8 27B

qwenLLM models262K context$0.90 / 1M input$6.40 / 1M output

Published catalog burn rate

Z AI: GLM 5.3

z-aiLLM models1M context$2.80 / 1M input$8.80 / 1M output

Published catalog burn rate

Z AI: GLM 5.3 Flash

z-aiLLM models1M context$0.15 / 1M input$0.50 / 1M output

Published catalog burn rate

QWEN: Qwen3.8 Flash

qwenLLM models1M context$0.32 / 1M input$0.94 / 1M output

Published catalog burn rate

OPENAI: GPT 4o

Tool callingStructured outputVision

GPT-4o ('o' for 'omni') is OpenAI's multimodal model supporting text and image inputs with text output. It matches GPT-4 Turbo's intelligence while being roughly twice as fast and more cost-effective, with strong creative writing, file understanding, multilingual performance, and vision capabilities.

Published catalog burn rate

openaiLLM models120K context$5.00 / 1M input$20.00 / 1M outputtext, image, file input

OPENAI: GPT 4

Tool callingStructured outputVision

GPT-4 Turbo is OpenAI's high-capability GPT-4 model with vision, supporting JSON mode and function calling on vision requests. Training data extends through December 2023.

Published catalog burn rate

openaiLLM models120K context$20.00 / 1M input$60.00 / 1M outputtext, image input

OPENAI: GPT 3

openaiLLM models16K context$1.00 / 1M input$3.00 / 1M outputtext input

GPT-3.5 Turbo is OpenAI's fast, cost-effective model for chat and traditional completion tasks. It understands and generates natural language and code, with training data through September 2021.

Published catalog burn rate

OPENAI: GPT 5 Mini

openaiLLM models400K context$0.50 / 1M input$4.00 / 1M outputtext, image, file input

GPT-5 Mini is a compact version of GPT-5 for lighter-weight reasoning tasks. It keeps GPT-5's instruction-following and safety tuning while offering lower latency and cost, and succeeds OpenAI's o4-mini.

Published catalog burn rate

OPENAI: GPT 5

openaiLLM models400K context$2.50 / 1M input$20.00 / 1M outputtext, image, file input

GPT-5 is one of OpenAI's most advanced models, optimized for complex tasks that require step-by-step reasoning, instruction following, and accuracy in high-stakes use cases. It delivers major gains in code quality and reasoning, with reduced hallucination and sycophancy and strong performance on coding, writing, and health-related tasks.

Published catalog burn rate

OPENAI: GPT 5.1

openaiLLM models400K context$2.50 / 1M input$20.00 / 1M outputtext, image, file input

GPT-5.1 is a frontier-grade model in the GPT-5 series with stronger general-purpose reasoning, improved instruction adherence, and a more natural conversational style. Its adaptive reasoning allocates compute dynamically, responding quickly to simple queries and going deeper on complex ones, delivering consistent gains across math, coding, and structured analysis with reliable tool use.

Published catalog burn rate

ANTHROPIC: Claude 3 Haiku

Tool callingVision

Claude 3 Haiku is Anthropic's fastest and most compact Claude 3 model, built for near-instant, targeted responses with multimodal input support.

Published catalog burn rate

anthropicLLM models100K context$0.50 / 1M input$2.50 / 1M outputtext, image input

ANTHROPIC: Claude Sonnet 4.5

anthropicLLM models1M context$6.00 / 1M input$30.00 / 1M outputtext, image, file input

Claude Sonnet 4.5 is Anthropic's most advanced Sonnet model, optimized for real-world agents and coding. It posts state-of-the-art results on coding benchmarks such as SWE-bench Verified and is built for extended autonomous operation, with improved tool orchestration, speculative parallel execution, and efficient context and memory management. It suits software engineering, cybersecurity, financial analysis, and research agents.

Published catalog burn rate

ANTHROPIC: Claude Opus 4.5

anthropicLLM models200K context$10.00 / 1M input$50.00 / 1M outputtext, image, file input

Claude Opus 4.5 is Anthropic's frontier reasoning model, optimized for complex software engineering, agentic workflows, and long-horizon computer use. It offers strong multimodal capabilities, competitive coding and reasoning performance, and improved robustness to prompt injection, with a verbosity control (low, medium, or high) to trade off speed, depth, and token usage. It supports advanced tool use, extended context management, and coordinated multi-agent setups for autonomous research, debugging, and multi-step planning.

Published catalog burn rate

GOOGLE: Gemini 2

Published catalog burn rate

googleLLM models120K context$0.20 / 1M input$0.80 / 1M output

GOOGLE: Gemini 2.5 Pro

googleLLM models1.1M context$2.50 / 1M input$20.00 / 1M outputtext, image, audio, video, file input

Gemini 2.5 Pro is Google's state-of-the-art model for advanced reasoning, coding, mathematics, and scientific tasks. Its 'thinking' capabilities let it reason through responses with greater accuracy and nuanced context handling, achieving top-tier results across benchmarks, including a first-place position on the LMArena leaderboard.

Published catalog burn rate

DEEPSEEK: DeepSeek Chat v3.1

deepseekLLM models164K context$0.40 / 1M input$1.60 / 1M outputtext input

DeepSeek-V3.1 is a large hybrid reasoning model (671B parameters, 37B active) supporting both thinking and non-thinking modes. It extends DeepSeek-V3 with two-phase long-context training up to 128K tokens and FP8 inference, improving tool use, code generation, and reasoning efficiency to a level comparable with DeepSeek-R1 on hard benchmarks while responding faster. It supports structured tool calling and code and search agents for research, coding, and agentic workflows.

Published catalog burn rate

TOGETHER: Llama 3.1 8B Instruct

togetherLLM models130K context$3.60 / 1M input$3.60 / 1M outputtext input

Llama 3.1 8B Instruct is the fast, efficient 8B instruction-tuned model in Meta's Llama 3.1 family, with strong performance against leading closed-source models in human evaluations.

Published catalog burn rate

TOGETHER: Llama 3.1 70B Instruct

togetherLLM models130K context$1.76 / 1M input$1.76 / 1M outputtext input

Llama 3.1 70B Instruct is the 70B instruction-tuned model in Meta's Llama 3.1 family, optimized for high-quality dialogue and competitive with leading closed-source models in human evaluations.

Published catalog burn rate

TOGETHER: Llama 3.1 405B Instruct

Published catalog burn rate

togetherLLM models130K context$7.00 / 1M input$7.00 / 1M output

TOGETHER: Mistral 7B Instruct

Published catalog burn rate

togetherLLM models32K context$0.40 / 1M input$0.40 / 1M output

TOGETHER: Mixtral 8x7B Instruct

Published catalog burn rate

togetherLLM models32K context$1.20 / 1M input$1.20 / 1M output

ANTHROPIC: Claude 3 Opus

anthropicLLM models100K context$26.00 / 1M input$74.00 / 1M output

Published catalog burn rate

ANTHROPIC: Claude 3 Sonnet

anthropicLLM models100K context$9.00 / 1M input$30.00 / 1M output

Published catalog burn rate

GOOGLE: Gemini 1.5

googleLLM models120K context$0.35 / 1M input$0.35 / 1M output

Published catalog burn rate

TOGETHER: Llama 2 Chat

togetherLLM models4K context$1.40 / 1M input$1.40 / 1M output

Published catalog burn rate

Z AI: GLM 4.6

z-aiLLM models128K context$0.78 / 1M input$3.80 / 1M outputtext input

GLM-4.6 expands the context window to 200K tokens and delivers higher coding benchmark scores and stronger real-world performance in coding tools, including more visually polished front-end generation. It improves reasoning with tool use during inference, performs better as a tool-using and search agent within agent frameworks, and aligns more naturally in writing and role-play.

Published catalog burn rate

QWEN: Qwen3 Coder

qwenLLM models128K context$0.44 / 1M input$3.60 / 1M outputtext input

Qwen3-Coder-480B-A35B-Instruct is a Mixture-of-Experts code generation model from the Qwen team, optimized for agentic coding tasks such as function calling, tool use, and long-context reasoning over repositories. It has 480 billion total parameters, with 35 billion active per forward pass (8 of 160 experts).

Published catalog burn rate

MOONSHOTAI: Kimi K2

Tool calling

Kimi K2 Instruct is a large-scale Mixture-of-Experts model from Moonshot AI with 1 trillion total parameters (32 billion active per forward pass), optimized for agentic capabilities including advanced tool use, reasoning, and code synthesis. It excels across coding (LiveCodeBench, SWE-bench), reasoning (ZebraLogic, GPQA), and tool-use benchmarks, and supports long-context inference up to 128K tokens.

Published catalog burn rate

moonshotaiLLM models128K context$1.14 / 1M input$4.60 / 1M outputtext input

ANTHROPIC: Claude Opus 4.8