OpenweightSuperintelligenceUnlocked.
The gap between open and closed models is disappearing.
The Numbers Don't Lie
Open-weight models now match and outperform proprietary models across intelligence, coding, and agentic benchmarks.
Claude Opus 4.8
- Intelligence
- 42.0
- Coding
- 74.3
- Agentic
- 42.6
GLM 5.3 Flash
- Intelligence
- 41.9
- Coding
- 71.5
- Agentic
- 51.2
GPT-5.6 Sol
- Intelligence
- 47.1
- Coding
- 77.4
- Agentic
- 50.5
DeepSeek V4.1 Flash
- Intelligence
- 39.5
- Coding
- 74.2*
- Agentic
- 54.8*
Grok 4.6
- Intelligence
- 44.4
- Coding
- 76.8
- Agentic
- 53.4
Qwen 3.8 2.4T
- Intelligence
- 40.0
- Coding
- 71.9
- Agentic
- 50.4
Same frontier. Different ownership. Results are shown as a directional comparison across representative dimensions; performance varies by evaluation, release, and configuration.
OPEN-SOURCE COMPUTE AT UNPRECEDENTED SCALE.
One API for curated open-weight modelswith providers, keys, and compute handled behind the scenes.
INFERENCE INFRASTRUCTURE.
BUILT FOR THE NEW ERA
Use a first-party Openweight API key to send text conversations to the open model gateway. Provider credentials stay behind the service.
/api/v1/chat/completionsBearer API keyAuthenticate with Authorization: Bearer <your key>.
model accepts auto or an active open model ID.
stream may be set to true for server-sent event responses.
{
"model": "auto",
"messages": [
{ "role": "user", "content": "Hello" }
]
}Messages support system, user, and assistant roles with text content.
Inference requires verified holder eligibility and available compute. Usage is settled from provider-reported usage.