OpenRouter(@OpenRouterAI)

Deepgram Nova-3: general-purpose speech-to-text with monolingual and multilingual transcription supp...

OpenRouter (@OpenRouter) — Thu, 23 Jul 2026 13:00:11 GMT

Deepgram Nova-3: general-purpose speech-to-text with monolingual and multilingual transcription support.

openrouter.ai/deepgram/nova-3

💬2🔄1❤️8👀2059📊3

Two audio models from @DeepgramAI are now live on OpenRouter. Nova-3 for speech-to-text. Aura-2 for...

OpenRouter (@OpenRouter) — Thu, 23 Jul 2026 13:00:08 GMT

Two audio models from @DeepgramAI are now live on OpenRouter.

Nova-3 for speech-to-text. Aura-2 for text-to-speech, with a multilingual voice catalog.

More on each below 🧵

💬7🔄3❤️31👀5842📊11

Gemini 3.5 Flash-Lite: Google’s fastest 3.5-series model, built for low-latency, high-throughput age...

OpenRouter (@OpenRouter) — Tue, 21 Jul 2026 21:54:25 GMT

Gemini 3.5 Flash-Lite: Google’s fastest 3.5-series model, built for low-latency, high-throughput agentic search, document processing, and subagents. Running on OpenRouter at 150+ tok/s

openrouter.ai/google/gemini-…

💬0🔄0❤️6👀3016📊1

Gemini 3.6 Flash: improved coding, knowledge work, multimodal performance, and agentic planning. Goo...

OpenRouter (@OpenRouter) — Tue, 21 Jul 2026 21:54:22 GMT

Gemini 3.6 Flash: improved coding, knowledge work, multimodal performance, and agentic planning. Google reports 17% fewer output tokens than 3.5 Flash.

openrouter.ai/google/gemini-…

💬5🔄1❤️10👀3438📊5

Launched today on OpenRouter: Gemini 3.6 Flash and Gemini 3.5 Flash-Lite! Both strong updates to th...

OpenRouter (@OpenRouter) — Tue, 21 Jul 2026 21:54:19 GMT

Launched today on OpenRouter: Gemini 3.6 Flash and Gemini 3.5 Flash-Lite!

Both strong updates to their model families, with high throughput (150+ tok/s) for agents, from token-efficient coding and knowledge work to low-latency, high-volume subagents.

Details below 🧵

💬7🔄11❤️167👀13229📊24

Or use it for free: https://t.co/BIpIDxumOr Free requests may be used by Poolside to train and impr...

OpenRouter (@OpenRouter) — Tue, 21 Jul 2026 18:19:50 GMT

Or use it for free:
openrouter.ai/poolside/lagun…

Free requests may be used by Poolside to train and improve its models.

💬0🔄2❤️13👀2870📊2

Laguna S 2.1 from @poolsideai is live on OpenRouter! An open-weight 118B-A8B MoE for agentic coding...

OpenRouter (@OpenRouter) — Tue, 21 Jul 2026 18:19:47 GMT

Laguna S 2.1 from @poolsideai is live on OpenRouter!

An open-weight 118B-A8B MoE for agentic coding and long-horizon work, with a 1M context window. Scores: 70.2% on Terminal-Bench 2.1 and 40.4% on DeepSWE.

Try it:
openrouter.ai/poolside/lagun…

💬8🔄10❤️192👀22360📊29

This past weekend, OpenRouter saved 22,000 users over $100,000 on just a single model. We're contin...

OpenRouter (@OpenRouter) — Tue, 21 Jul 2026 16:26:46 GMT

This past weekend, OpenRouter saved 22,000 users over $100,000 on just a single model.

We're continuously sourcing the best prices in the market for open-weight models, and it will get even better very soon!

Shashank Goyal@shashankgoyal95

OpenRouter is the best way to source inference. Period. 

Over the weekend we saved 22000 users over $100,000 on GLM 5.2 because providers gave discounts during low demand periods

Users had to do nothing, they just got way better prices! Over a 50% drop from weekday pricing.

💬12🔄11❤️325👀25656📊36

Krea 2 Medium Turbo: a distilled, speed-focused variant for rapid iteration and graphic design explo...

OpenRouter (@OpenRouter) — Mon, 20 Jul 2026 19:39:40 GMT

Krea 2 Medium Turbo: a distilled, speed-focused variant for rapid iteration and graphic design exploration.

openrouter.ai/krea/krea-2-me…

💬0🔄0❤️2👀2373

Krea 2 Medium: the faster, cost-efficient starting point for broad use. Stable, consistent generatio...

OpenRouter (@OpenRouter) — Mon, 20 Jul 2026 19:39:37 GMT

Krea 2 Medium: the faster, cost-efficient starting point for broad use. Stable, consistent generations across illustration, anime, painting, and expressive styles.

openrouter.ai/krea/krea-2-me…

💬2🔄0❤️6👀3032📊2

Live on OpenRouter: Krea 2 from @krea_ai. Three image models built for creative control, style refe...

OpenRouter (@OpenRouter) — Mon, 20 Jul 2026 19:39:32 GMT

Live on OpenRouter: Krea 2 from @krea_ai.

Three image models built for creative control, style references, and rapid iteration.

More on Krea 2 Large, Medium, and Medium Turbo below 🧵

💬7🔄5❤️52👀8906📊13

Grok Imagine Video 1.5 from @SpaceXAI is now live on OpenRouter! Bring any still image to life with...

OpenRouter (@OpenRouter) — Mon, 20 Jul 2026 13:43:51 GMT

Grok Imagine Video 1.5 from @SpaceXAI is now live on OpenRouter!

Bring any still image to life with an optional text prompt. Shape the motion and camera work. Set the atmosphere and physics with built-in sound and dialogue.

Try it: openrouter.ai/x-ai/grok-imag…

💬16🔄9❤️145👀11400📊29

9 great new model launches this week, including Kimi K3, Muse Spark, and Thinking Machines Inkling. ...

OpenRouter (@OpenRouter) — Sat, 18 Jul 2026 15:06:44 GMT

9 great new model launches this week, including Kimi K3, Muse Spark, and Thinking Machines Inkling.

Explore them here: openrouter.ai/models

💬9🔄8❤️218👀16100📊31

Inkling from @thinkymachines is live on OpenRouter. An open-weights MoE model with 975B total / 41B...

OpenRouter (@OpenRouter) — Fri, 17 Jul 2026 22:22:09 GMT

Inkling from @thinkymachines is live on OpenRouter.

An open-weights MoE model with 975B total / 41B active parameters, 1M context, and controllable reasoning across text, images, and audio.

Try it: openrouter.ai/thinkingmachin…

💬7🔄8❤️180👀15119📊27

Muse Spark 1.1 is available now on OpenRouter for US-based developers. https://t.co/aoegJZMG2w

OpenRouter (@OpenRouter) — Thu, 16 Jul 2026 17:15:08 GMT

Muse Spark 1.1 is available now on OpenRouter for US-based developers.

openrouter.ai/meta/muse-spar…

💬1🔄2❤️10👀2862📊3

Muse Spark 1.1 is a multimodal reasoning model designed for agentic tasks, including coding, tool us...

OpenRouter (@OpenRouter) — Thu, 16 Jul 2026 17:15:05 GMT

Muse Spark 1.1 is a multimodal reasoning model designed for agentic tasks, including coding, tool use, computer use, and multimodal understanding. See how it performs across these capabilities:

💬2🔄2❤️14👀3357📊4

Muse Spark 1.1 from @AIatMeta is live on OpenRouter for US-based devs! At $1.25/M in and $4.25/M ou...

OpenRouter (@OpenRouter) — Thu, 16 Jul 2026 17:15:01 GMT

Muse Spark 1.1 from @AIatMeta is live on OpenRouter for US-based devs!

At $1.25/M in and $4.25/M out, the model is poised to be one of the most price-efficient, high-intelligence models on OpenRouter. Built for production agentic workloads w/ multimodal inputs, viable at scale.

💬20🔄14❤️194👀21309📊42

Easily compare all GPT 5.6 variants side by side: https://t.co/TMNRzLWmwg

OpenRouter (@OpenRouter) — Wed, 15 Jul 2026 15:18:16 GMT

Easily compare all GPT 5.6 variants side by side: openrouter.ai/chat?models=op…

💬7🔄4❤️83👀10670📊14

Plus: docs search now runs on @mintlify, read-only tools no longer trigger permission prompts, and e...

OpenRouter (@OpenRouter) — Tue, 14 Jul 2026 20:43:00 GMT

Plus: docs search now runs on @mintlify, read-only tools no longer trigger permission prompts, and every tool renamed to a clean verb-noun format. Get started in 30 seconds: openrouter.ai/mcp

💬0🔄1❤️17👀8597📊3

Two new tools: generate-image (works on Desktop apps that can view images, billed to your key) and s...

OpenRouter (@OpenRouter) — Tue, 14 Jul 2026 20:42:58 GMT

Two new tools: generate-image (works on Desktop apps that can view images, billed to your key) and send-feedback (report a bad generation straight to the OpenRouter team, no support form).

💬1🔄0❤️3👀3270📊1

We just shipped a number of updates to the OpenRouter MCP - 2 weeks of OpenRouter MCP updates in one...

OpenRouter (@OpenRouter) — Tue, 14 Jul 2026 20:42:46 GMT

We just shipped a number of updates to the OpenRouter MCP - 2 weeks of OpenRouter MCP updates in one thread + demo.

Your agent can now discover, rank, test, and report on models without leaving your editor 🧵

Your browser does not support the video tag.

💬17🔄16❤️223👀20640📊48

New: separate leaderboards for open-weight models and closed-weight models on https://t.co/jy765ILVB...

OpenRouter (@OpenRouter) — Tue, 14 Jul 2026 03:05:45 GMT

New: separate leaderboards for open-weight models and closed-weight models on openrouter.ai/rankings

Your browser does not support the video tag.

💬10🔄15❤️169👀15951📊29

Learn how to use service tiers: https://t.co/pu0l9JCpgD

OpenRouter (@OpenRouter) — Mon, 13 Jul 2026 17:50:21 GMT

Learn how to use service tiers:
openrouter.ai/docs/guides/fe…

💬1🔄0❤️10👀3983📊3

Service tiers are now first-class on OpenRouter. See tier-specific latency and throughput in `/v1/m...

OpenRouter (@OpenRouter) — Mon, 13 Jul 2026 17:50:18 GMT

Service tiers are now first-class on OpenRouter.

See tier-specific latency and throughput in `/v1/models/endpoints` so you know exactly what you’re paying for.

Route with copyable slugs like `openai/flex` and `xai/priority`, or keep using the existing `service_tier` param.

💬7🔄3❤️164👀14522📊25

Read our story about the rebrand: https://t.co/fZ80awH7eX

OpenRouter (@OpenRouter) — Mon, 13 Jul 2026 15:49:26 GMT

Read our story about the rebrand: openrouter.ai/blog/announcem…

💬2🔄2❤️45👀6649📊7

OpenRouter has a new look. As we’ve grown into the infrastructure powering the intelligence era, ou...

OpenRouter (@OpenRouter) — Mon, 13 Jul 2026 15:49:23 GMT

OpenRouter has a new look.

As we’ve grown into the infrastructure powering the intelligence era, our brand needed to evolve too.

Built on clarity, precision, and purpose.

For developers. For enterprises. For everyone turning intelligence into impact.

💬167🔄47❤️1259👀103168📊284

GPT-5.6 Luna: the fast, low-cost tier. Tuned for high-volume, latency-sensitive work like chat, clas...

OpenRouter (@OpenRouter) — Thu, 09 Jul 2026 17:32:19 GMT

GPT-5.6 Luna: the fast, low-cost tier. Tuned for high-volume, latency-sensitive work like chat, classification, and lightweight agents, with real GPT-5.6 reasoning.

openrouter.ai/openai/gpt-5.6…

💬1🔄0❤️13👀4202📊2

GPT-5.6 Terra: the balanced everyday driver. Coding, reasoning, and agentic tasks at GPT-5.5-competi...

OpenRouter (@OpenRouter) — Thu, 09 Jul 2026 17:32:16 GMT

GPT-5.6 Terra: the balanced everyday driver. Coding, reasoning, and agentic tasks at GPT-5.5-competitive quality for roughly 2x lower cost.

openrouter.ai/openai/gpt-5.6…

💬2🔄2❤️18👀5303📊4

GPT-5.6 is live on OpenRouter! Sol (flagship), Terra (balanced), and Luna (fast + low-cost). In ou...

OpenRouter (@OpenRouter) — Thu, 09 Jul 2026 17:32:10 GMT

GPT-5.6 is live on OpenRouter!

Sol (flagship), Terra (balanced), and Luna (fast + low-cost).

In our early testing, the 5.6 models were more token efficient, leading to lower cost and faster time to task completion. AA's results agree.

More on Sol, Terra, and Luna below 🧵

💬25🔄42❤️585👀43704📊92

Use it now: https://t.co/yrkHFKgtBK

OpenRouter (@OpenRouter) — Wed, 08 Jul 2026 18:25:41 GMT

Use it now: openrouter.ai/x-ai/grok-4.5

💬1🔄1❤️10👀3867📊2

Grok 4.5 is live now on OpenRouter! Cursor’s contributions have helped make @SpaceXAI’s model fron...

OpenRouter (@OpenRouter) — Wed, 08 Jul 2026 18:25:40 GMT

Grok 4.5 is live now on OpenRouter! 

Cursor’s contributions have helped make @SpaceXAI’s model frontier-level on long-horizon engineering tasks.

SpaceXAI@SpaceXAI

Announcing Grok 4.5, our first model trained specifically for coding and agents. It was trained with Cursor and offers frontier intelligence at leading speeds and cost efficiency.

x.ai/news/grok-4-5

Your browser does not support the video tag.

💬22🔄20❤️310👀18117📊41

Use Grok 4.5 from @SpaceXAI the moment it goes live, with no code change. ~x-ai/grok-latest always ...

OpenRouter (@OpenRouter) — Wed, 08 Jul 2026 15:42:40 GMT

Use Grok 4.5 from @SpaceXAI the moment it goes live, with no code change.

~x-ai/grok-latest always routes to the newest Grok. Point at it now (Grok 4.3 today) and it auto-upgrades to 4.5 the second it ships.

Try it: openrouter.ai/~x-ai/grok-lat…

💬21🔄32❤️545👀35917📊77

New: one-click ZDR (zero-data retention) in the Chatroom Compare models side-by-side with full priv...

OpenRouter (@OpenRouter) — Wed, 08 Jul 2026 14:57:23 GMT

New: one-click ZDR (zero-data retention) in the Chatroom

Compare models side-by-side with full privacy: openrouter.ai/chat

💬2🔄4❤️84👀9173📊11

So what actually moves the bill? Reasoning effort, by a wide margin. Detail swung accuracy 2 to 17 ...

OpenRouter (@OpenRouter) — Tue, 07 Jul 2026 21:08:35 GMT

So what actually moves the bill? Reasoning effort, by a wide margin.

Detail swung accuracy 2 to 17 points while barely touching cost, while capping reasoning effort cut cost 50 to 75 percent with accuracy moving 1 to 2 points, within noise.

💬1🔄1❤️10👀3291📊2

TLDR: Reasoning models → keep detail on auto or high, control cost with reasoning effort Non-reaso...

OpenRouter (@OpenRouter) — Tue, 07 Jul 2026 21:08:35 GMT

TLDR:

Reasoning models → keep detail on auto or high, control cost with reasoning effort

Non-reasoning models → low detail genuinely saves money and latency if you can absorb the accuracy hit

Full benchmark with per-model tables and methodology: openrouter.ai/blog/insights/…

💬0🔄0❤️14👀2965📊3

We benchmarked 1,730 visual reasoning questions across 5 models to test a common cost-saving trick: ...

OpenRouter (@OpenRouter) — Tue, 07 Jul 2026 21:08:33 GMT

We benchmarked 1,730 visual reasoning questions across 5 models to test a common cost-saving trick: setting image detail to "low."

Surprise finding: it often backfires. The model burns extra reasoning tokens squinting at a blurry image, and overall cost increases 👇

💬4🔄16❤️223👀32403📊32

V4 also broke DeepSeek away from the other Chinese models. At 6.6T weekly tokens it now runs well a...

OpenRouter (@OpenRouter) — Mon, 06 Jul 2026 15:15:20 GMT

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.

💬3🔄2❤️19👀4824📊5

Full breakdown across 457 trillion tokens of request data here: https://t.co/sV83QDrRCI

OpenRouter (@OpenRouter) — Mon, 06 Jul 2026 15:15:20 GMT

Full breakdown across 457 trillion tokens of request data here: openrouter.ai/blog/insights/…

💬0🔄0❤️11👀3272📊1

DeepSeek doubled its token share on OpenRouter in six months, going from 9% in January to 18% by Jun...

OpenRouter (@OpenRouter) — Mon, 06 Jul 2026 15:15:19 GMT

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 👇

💬17🔄33❤️420👀28702📊67

More about the MCP: https://t.co/1O8czSPLGz

OpenRouter (@OpenRouter) — Sun, 05 Jul 2026 16:49:35 GMT

More about the MCP: x.com/OpenRouter/sta…

OpenRouter@OpenRouter

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:

Your browser does not support the video tag.

💬2🔄0❤️12👀9996📊3

Get started here: https://t.co/PYkilNr1At

OpenRouter (@OpenRouter) — Sun, 05 Jul 2026 16:49:35 GMT

Get started here: openrouter.ai/docs/guides/ov…

💬0🔄0❤️14👀5085📊3

Saving 24x on inference costs with comparable quality, using the OpenRouter MCP:

OpenRouter (@OpenRouter) — Sun, 05 Jul 2026 16:49:34 GMT

Saving 24x on inference costs with comparable quality, using the OpenRouter MCP:

💬16🔄33❤️500👀48551📊111

Rube Goldberg machine built with Fable 5, for $4.44 👀 Prompt available here: https://t.co/zlLXo01J...

OpenRouter (@OpenRouter) — Thu, 02 Jul 2026 17:46:31 GMT

Rube Goldberg machine built with Fable 5, for $4.44 👀

Prompt available here:reddit.com/r/openrouter/c…M

Your browser does not support the video tag.

💬8🔄2❤️86👀11877📊17

Try it at https://t.co/Q4Ql4Dp0qX

OpenRouter (@OpenRouter) — Wed, 01 Jul 2026 19:44:30 GMT

Try it at openrouter.ai/anthropic/clau…

💬0🔄0❤️8👀3985📊1

Claude Fable 5 from @Anthropic is back on OpenRouter! Anthropic is redeploying it globally with new...

OpenRouter (@OpenRouter) — Wed, 01 Jul 2026 19:44:27 GMT

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.

💬34🔄25❤️502👀29133📊77

Open-weight provider meta-benchmarking, run and published continuously by OpenRouter. Powers our to...

OpenRouter (@OpenRouter) — Tue, 30 Jun 2026 23:18:54 GMT

Open-weight provider meta-benchmarking, run and published continuously by OpenRouter.

Powers our tool-call routing and ensures you receive the highest-quality inference: x.com/OpenRouter/sta…

OpenRouter@OpenRouter

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…

💬0🔄0❤️2👀4025

@AnthropicAI Always stay on the latest Sonnet by using ~anthropic/claude-sonnet-latest in your API c...

OpenRouter (@OpenRouter) — Tue, 30 Jun 2026 18:14:48 GMT

@AnthropicAI 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!

💬1🔄0❤️4👀2685📊1