One API for 250+ LLMs. Routing, failover, budgets, guardrails and audit logs for teams. SOC 2 Type II. Cloud or self-hosted: https://xitter.cf/t.co/KLJct0BnZX
World
Joined June 2025
- Tweets1K
- Following25
- Followers1.6K
- Likes683
Pinned Tweet
A typo fix and a concurrency proof shouldn’t need the same model.
Smart Routing in LLM Gateway picks from your model pool based on task difficulty. Just set your model to "smart".
Watch it work in @opencode, @Empryo_ai , devpass-code, and The Lounge 👇
LLM Gateway retweeted
My Samsung TV now boots into an Apple TV style launcher instead of Smart Hub. An AI agent built it in 2 days.
2,306 calls, $173.85. Opus 5.5 built it; DeepSeek V4.1 Flash did 893 cheap lookups via @llmgateway for $3.53.
Story, screenshots, DIY prompt:
smakosh.com/blog/replacing-s…
LLM Gateway retweeted
Both Kimi K3.1 Pro and K3.1 Pro High speed are live on @llmgateway
- llmgateway.io/models/kimi-k3…
- llmgateway.io/models/kimi-k3… (over 111.3 t/s)
Introducing Kimi K3.1 Pro: our most capable 3T-class open-weight model.
🔹 1M-token context, 2.8T total / 104B active params (MoE), Multimodal.
🔹 Full-project coding: games in Unreal, Unity & Godot.
🔹 Excels at parametric 3D modeling in Blender & CAD.
🔹 More reliable at finance work: Accurate modeling with hallucination down by 65%.
🔹 Matching the strongest proprietary models across coding, agentic and knowledge benchmarks.
Kimi K3.1 Pro and K3.1 Pro Highspeed (2–3× faster) are now live on Kimi.ai, Kimi Work, Kimi Code, and the Kimi API.
Open weights by Nov 4, 2026.
🔗 API: platform.kimi.ai
🔗 Tech blog: kimi.ai/blog/kimi-k3-1-pro
Security teams used to ask "are you SOC 2?"
Now they ask "you use AI to write code. How do you keep it secure?"
Our answer, with the real PRs: advisories patched in days, a monthly AI security audit, AI + human review on every PR, keys that fail closed.
llmgateway.io/blog/how-we-ke…
LiteLLM was compromised again.
litellm 1.107.0 on PyPI and its Docker image (incl. latest and main-stable) shipped an infostealer and a backdoor that dumped every provider key in the proxy's DB.
How to check safely, what to rotate, and how to move off:
llmgateway.io/blog/litellm-c…
LLM Gateway is now fully available and open source thanks to Claude
i had 1000 Claude agents running in parallel for 69 hours and i managed to reverse engineer @llmgateway
i'm making it available here so that you can save your tokens - 100% free & open-source:
github.com/theopenco/llmgate…
API keys pile up.
Old CI jobs, finished demos, teammates who moved on.
Their keys still authenticate, and until now cleaning them up meant opening every project and deactivating keys one at a time.
We just shipped bulk API key management in LLM Gateway:
→ Every key from every project in one table, filterable by project, status and usage
→ Usage for the last 24 hours, 7, 30 or 90 days, plus a Last Used column
→ Amber and red badges for keys idle 30+ and 90+ days, and a banner that selects them in one click
→ Bulk deactivate, reactivate or delete, with a confirmation and a report of any keys it skipped
It's live for everyone now: open Organization → API Keys in your dashboard.
Changelog: llmgateway.io/changelog/api-…
LLM Gateway retweeted
Released yesterday. Already #2.
DeepSeek V4.1 Flash 1017 moved 11.7B tokens through LLM Gateway in the last 24h, 26.3% of the board. The original V4.1 Flash kept #1 at 22.6B, down just 1.1%.
Two DeepSeek snapshots, 77.0% of 44.5B tokens.
Switching cost: one model string.
ALT Stacked bar chart of LLM Gateway token volume by model over the last 24 hours: 44.5B tokens, 729.1k requests, 140 models ranked; DeepSeek V4.1 Flash leads with 50.7% of tokens, followed by DeepSeek V4.1 Flash 1017 at 26.3% and GLM-5.3 Flash at 4.8%.
🚀 DeepSeek-V4.1-Flash-1017 is here!
Stronger benchmark performance, faster task completion, and shorter outputs than the previous version—a Pareto improvement in performance and cost. Now live on our website, app, and API.
🛠️ Optimized for @DeepSeekHarness, Tencent CodeBuddy, WorkBuddy, and more. Upgrade to DeepSeek Harness v0.2.1-rc.1 or later for the best experience with the new model.
1/4
Provider keys on LLM Gateway can now be capped in tokens, not just requests.
Set RPM, RPD, TPM or TPD on each key to match your upstream quota. When a key hits its cap, the next key or provider takes the request instead of returning a 429.
docs.llmgateway.io/learn/pro…
DevPass Code 2.0 is out.
The Smart Router picks the model for every message: a decision model reads your request and sends it to Claude Haiku, Sonnet or Opus.
You choose who decides: GPT-6 Luna Decisions, Microsoft-Decision-1 or Jev.
Plus session tabs.
npm i -g devpass-code
How it works, pricing and shortcuts: llmgateway.io/blog/devpass-c…
LLM Gateway retweeted
GPT-6 Luna now answers through LLM Gateway's new OpenAI-compatible Decisions endpoint: typed yes/no, choice and score answers with probabilities, not text. 1.05M context, vision input, from $0.10/1M input tokens. llmgateway.io/models/gpt-6-l…
LLM Gateway retweeted
Microsoft's first model lands on LLM Gateway: Decision-1, built for classification, routing, ranking and scoring via the new /v1/decisions endpoint. From $0.042/1M input tokens on Azure. llmgateway.io/models/microso…
LLM Gateway retweeted
260.5B tokens. 6.2M requests. 201 models. One week through LLM Gateway.
DeepSeek V4.1 Flash took 41% of every token on its own, up 46.9% week over week.
Muse Spark 1.3 Contributor grew 353% and sits third overall.
Seven days is long enough that this isn't noise.
OpenAI's Decisions API is live on LLM Gateway.
/v1/decisions returns typed answers with probabilities, not generated text. GPT-6 Luna Decisions reads text and images at $0.10/1M input tokens. Microsoft-Decision-1 and Jev run on the same request shape.
llmgateway.io/blog/openai-de…
Changelog, pricing table and the /v1/systemone migration guide:
llmgateway.io/changelog/open…
Perplexity's Search API is live on LLM Gateway.
/v1/search returns ranked web results with page snippets, not a model-written answer. $5 per 1K searches, $1 in fast mode, and failed requests are free.
llmgateway.io/changelog/perp…
LLM Gateway retweeted
Grok Imagine Video 1.5 Lite is live on LLM Gateway. xAI's cheapest video model yet: text-to-video at $0.02/sec (480p), with audio, no input image needed. llmgateway.io/models/grok-im…
Haiku 5.5 is live now llmgateway.io/models/claude-…
