Skip to main content

💫 Cosmic Routers

While technically “routers” and not “models” - 💫 Cosmic Routers act through our OpenAI and Anthropic compatible API as if they were models themselves. They intelligently route every request to one of their underlying models.
💫 Orbit
costa/orbit
SWEBench: 57.9
Description: Orbit always includes current best practices in intelligent model routing - it is Costa’s “frontier” model router. Orbit uses a mix of routing strategies, including semantic task analysis, performance-based benchmarking, and adaptive load balancing. It is designed to minimize manual intervention. It is particularly good at architecting projects and writing code in Python, C++, Javascript, Go and Ruby (although it does well in other languages).

Models Included:

    anthropic/claude-sonnet-4.5

    anthropic/claude-sonnet-4

    anthropic/claude-3.7-sonnet

    anthropic/claude-3.5-haiku

    anthropic/claude-haiku-4.5

    openai/gpt-5-mini

    openai/gpt-4.1

    google/gemini-2.5-flash

    google/gemini-2.5-pro

    qwen/qwen3-coder

    moonshotai/kimi-k2

Stability: 🚀 Active Tuning

May Add Models: ✅ Yes

Last Changed 2025-10-16
💫 Quasar
costa/quasar
SWEBench: 56.2
Description: Quasar is a stable model router based on Avengers Pro routing logic. It trades performance for time efficiency in an extremely clever manner. Expect fewer updates to Quasar as it is relatively stable

Models Included:

    anthropic/claude-sonnet-4

    anthropic/claude-3.5-haiku

    qwen/qwen3-coder

    moonshotai/kimi-k2

Stability: 🛰️ Stable

May Add Models: 🛑 No

Last Changed 2025-08-23

Other Top Models

Costa provides access to almost 100 models - here are the top ones used for coding:
GPT 4.1
gpt-4.1
SWEBench: 39.6
Provider:OpenAI
Claude 3.7 Sonnet
claude-sonnet-3.7
SWEBench: 52.8
Provider:Anthropic
Claude 4 Sonnet
claude-sonnet-4
SWEBench: 64.9
Provider:Anthropic
Meta Llama 4 Maverick 17b
llama-4-maverick-17b-128e-instruct
SWEBench: 21
Provider:Meta (via Groq/Google Vertex)
Gemini 2.5 Pro
gemini-2.5-pro
SWEBench: 53.6
Provider:Google
Gemini Flash 2.5
gemini-2.5-flash
SWEBench: 28.7
Provider:Google
Claude Haiku 4.5
claude-haiku-4.5
Provider:Anthropic
Qwen3 Coder
qwen3-coder
SWEBench: 55.4
Provider:Qwen (via Groq/Google Vertex)
O4 Mini
o4-mini
SWEBench: 45
Provider:OpenAI
Qwen3 32b
qwen3-32b
SWEBench: 42.2
Provider:Qwen (via Groq/Google Vertex)
Kimi K2
kimi-k2
SWEBench: 43.8
Provider:MoonshotAI (via Groq/Google Vertex)
GPT OSS 20b
gpt-oss-20b
SWEBench: 5
Provider:OpenAI (via Groq/Google Vertex)
Qwen 2.5 Coder
qwen-2.5-coder
SWEBench: 9
Provider:Qwen (via Groq/Google Vertex)
… and many more. If you are logged in to Costa, you can see all of the models we provide access to from your Models Page.