Models available on Felo

Anthropic's Haiku-class route, succeeding Claude Haiku 4.5 as the fast tier of the Claude 5.5 family. It is built for high-volume and latency-sensitive workloads such as support bots, classification, extraction, and search-backed answers, and pairs a 1M-token context window with cache read and cache-write rates that keep high-throughput loops affordable.

An upgrade to GPT-6 Sol, still positioned below the flagship GPT-6 Astra. It handles agentic coding, computer use, document-heavy professional work, and multi-step workflow automation, approaching Astra-level results on those tasks at a much lower cost, with fewer factual errors and closer adherence to explicit restrictions than GPT-6 Sol. Cache reads are billed at half of GPT-6 Sol's rate.

The same underlying model as GPT-6.1 Sol, served with reasoning mode set to pro for higher-quality responses on complex tasks. Choose it when accuracy matters more than latency.

Anthropic's Sonnet-class route for well-scoped everyday work, succeeding Claude Sonnet 5 as a direct upgrade. It is strongest at building features and fixing bugs, and pairs a 1M-token context window with cache read and cache-write rates that keep high-volume agent loops affordable.

Anthropic's flagship route for multi-step changes in large codebases, long-horizon agentic work, and demanding reasoning. It pairs a 1M-token context window with cached input and cache-write rates for production agent loops.

The fast, cost-efficient tier of the GPT-6 series, positioned below GPT-6 Sol. It is suited for high-volume and latency-sensitive workloads such as chat, classification, and lightweight agentic loops.

The same underlying model as GPT-6 Luna, served with reasoning mode set to pro for higher-quality responses on complex tasks. It keeps Luna's low token pricing while trading latency for accuracy.

The cost-efficient high-end tier of the GPT-6 series, positioned below the flagship GPT-6 Astra and above the fast GPT-6 Luna route. It is built for demanding professional workloads that need strong reasoning without flagship pricing.

The same underlying model as GPT-6 Sol, served with reasoning mode set to pro for higher-quality responses on complex tasks. Choose it when accuracy matters more than latency.

The flagship Grok route for coding, agentic tasks, and knowledge work. It is particularly strong at long-running software engineering tasks, verifying its own work, and large-context analysis, with text, image, and file input.

Xiaomi's flagship foundation model, built at a scale of over 1T parameters to push the ceiling of capability on the most demanding tasks. It accepts text, image, audio, and video input with a 1M-token context window.

A sparse mixture-of-experts model from DeepSeek and the first built on the company's Causal Encoder-Decoder architecture. It activates a small fraction of parameters per token, keeping long-context reasoning fast and inexpensive.

OpenAI's most capable model, built for the hardest end-to-end work: complex reasoning, agentic coding, computer use, web research, and professional document creation. 1.05M context window with cached input and cache-write rates.

Z.AI's open-weight Flash route with a 1M-token context, native text, image, and video input, and hybrid sparse attention that keeps long-context accuracy while cutting compute cost.

A stealth frontier reasoning route with a 1M-token context and text, image, and video input, built for long-horizon coding, complex reasoning, and production agent workflows.

The most affordable Seedance route, supporting text-to-video, first and last frame control, and multimodal reference-to-video with image, video, and audio inputs. It renders 480p and 720p clips between 4 and 15 seconds.

ByteDance's newest Seedance route, built for long-form storytelling, multimodal reference-based generation, video editing, and video extension. It accepts up to 50 image, video, and audio reference assets, renders clips up to 30 seconds, and supports multilingual audiovisual generation.

The efficient GPT-5.6 tier for fast coding loops, chat, extraction, search synthesis, and high-volume agent workflows.

The flagship GPT-5.6 tier for complex reasoning, advanced coding, long-context analysis, and high-capability agent workflows.

The balanced GPT-5.6 tier for production coding agents, reasoning tasks, tool use, and multi-step app workflows.

A high-capability model route for research, analysis, coding assistance, and agent workflows that benefit from a large context window.

A strong default for modern applications that need long context, tool use, structured output, and reliable streaming responses.

A premium route for complex planning, multi-step analysis, high-stakes content generation, and advanced software engineering assistance.

A cost-efficient DeepSeek route for agent loops, coding, structured output, and tool use with a 1M-token context window and selectable thinking modes.

A strong general-purpose model route for coding, reasoning, research, and production agent workloads through Responses and Chat Completions compatible APIs.

A premium model route for deep reasoning, complex software engineering, long-horizon planning, and high-value agent workflows.

Xiaomi's flagship text route, delivering strong performance in general agentic capabilities, complex software engineering, and long-horizon tasks, with top rankings on benchmarks such as ClawEval, GDPVal, and SWE-bench Pro.

ByteDance's flagship Seedance route for text-to-video, first and last frame control, and multimodal reference-to-video. It is particularly strong at preserving character consistency, visual style, and camera movement from reference material, and it is the only Seedance route here that renders up to 4K.

A faster, lower-cost Seedance route for text-to-video, first and last frame control, and multimodal reference-to-video. It prioritizes generation speed and cost over maximum output quality, and renders up to 720p.

A general-purpose flagship route for agent workflows, long document understanding, coding assistance, and production-grade reasoning.

A compact, responsive model route for support bots, extraction, classification, and search-backed answers where latency matters.