every model choice is a bet.
it shouldn't be a blind one.
750T+ tokens routed across every major model and provider — what actually moves cost, latency, and reliability.
live at the top of salesforce tower w/
@deedydas (menlo) + @shashankgoyal95 (openrouter)
rsvp 👇
DeepSeek doubled its token share on OpenRouter in six months, going from 9% in January to 18% by June.
The driver is agentic work. New data deep dive in our blog 👇
V4 also broke DeepSeek away from the other Chinese models.
At 6.6T weekly tokens it now runs well ahead of Xiaomi (4.0T), Minimax (3.6T), and Tencent (3.8T), and it edges out Anthropic's 6.1T as the highest-token-volume model lab on the platform.
Introducing the OpenRouter MCP, live model intelligence right inside your agent
Your agent builds and ships, but when it comes to choosing the right model for the right job, it guesses from 6 month old training data
Watch it pick, price, and test the right model:
Claude Fable 5 from @Anthropic is back on OpenRouter!
Anthropic is redeploying it globally with new safeguards for cybersecurity misuse. Some coding and debugging requests may temporarily fall back to Opus 4.8 while classifiers are refined.
This month is, unsurprisingly, Cost Reduction Month.
In our data from the last 3 yrs, we commonly see major cost crunches right after the latest breakthrough.
We'll ship major features to help you cut inference costs at least once a week, starting with today.
Running list 👇
New server tool: Advisor
Let smaller models consult a higher-intelligence "advisor" model.
Helps them escape doom loops, and helps you migrate to cheaper models! 🧵
Official OpenRouter MCP: help your agents pull live data about model pricing, benchmarks, popularity, and more!
Avoid the accidental bad model slug being written into your codebase.
Introducing the OpenRouter MCP, live model intelligence right inside your agent
Your agent builds and ships, but when it comes to choosing the right model for the right job, it guesses from 6 month old training data
Watch it pick, price, and test the right model:
Open-weight provider meta-benchmarking, run and published continuously by OpenRouter.
Powers our tool-call routing and ensures you receive the highest-quality inference:
Tip: OpenRouter continuously runs GPQA and TAU-Bench on most open-weight models and publishes the results publicly.
This informs our AutoExacto meta-benchmark, used by default when routing tool calls.
Here, @parasail_io and @Zai_org rank first: openrouter.ai/z-ai/glm-5.2#p…
Always stay on the latest Sonnet by using ~anthropic/claude-sonnet-latest in your API calls.
~anthropic/claude-sonnet-latest now routes to Claude Sonnet 5!
Claude Sonnet 5 is rolling out on OpenRouter with a promo price: $2/M in and $10/M out!
It boosts agentic coding and pro workflows w/ flagship intelligence at Sonnet pricing. In early tests, agents were more reliable, faster, and easier to trust with larger tasks than 4.6.
Introducing Claude Sonnet 5, our most agentic Sonnet yet.
It makes plans, uses tools like browsers and terminals, and runs autonomously at a level that just a few months ago required larger and more expensive models.
Tip: OpenRouter continuously runs GPQA and TAU-Bench on most open-weight models and publishes the results publicly.
This informs our AutoExacto meta-benchmark, used by default when routing tool calls.
Here, @parasail_io and @Zai_org rank first: openrouter.ai/z-ai/glm-5.2#p…