OpenRouter's June 2026 insights post highlights four open-weight models it considers most important: DeepSeek V4 Flash, GLM 5.2, MiniMax M3, and NVIDIA Nemotron 3 Ultra. The piece argues that open-weight models have kept a consistent 3–6 month capability gap behind US frontier labs for roughly 18 months, and that shift DeepSeek V4 Flash is the lead recommendation: an MIT-licensed 284B-parameter / 13B-active mixture-of-experts model with a 1M-token context window that scores 79.0% on SWE-bench Verified within 1.6 points of the larger V4 Pro (80.6%). DeepSeek's first-party API pricing is $0.14 per million input tokens and $0.28 per mil

OpenRouter has raised a $113M Series B led by CapitalG, with participation from NVentures, ServiceNow Ventures, MongoDB Ventures, Snowflake Ventures, Databricks

OpenRouter announced a $113M Series B funding round led by CapitalG, with participation from NVentures, ServiceNow Ventures, MongoDB Ventures, Snowflake Ventures, Databricks Ventures, AMP PBC, and Pace Capital, alongside existing investors Andreessen Horowitz and Menlo Ventures. The round, dated May 28, 2026, provides In a separate technical post (June 12, 2026), OpenRouter detailed its Fusion approach, where a panel of budget models combined through OpenRouter's infrastructure outperformed GPT-5.5 and Claude Opus 4.8 on 100 complex research tasks. The blog also highlights recent product launches including the OpenRouter MCP Server

On August 21, 2026, OpenRouter launched Visual Image Benchmarks, a new evaluation suite that runs 39 image models (as of August 2026) through a curated set of 15 challenging prompts and presents every result in a sortable grid. The prompts are grouped into seven capability families — improbable scenes, counting, text r Releasebot's coverage frames the launch as filling a gap left by LLM-as-a-judge evals and curated samples, and notes that OpenRouter hinted additional modalities (video and audio) will get similar visual evaluations next. The release is positioned as a developer-facing decision tool for picking image models on OpenRout

Contrary Research published a business breakdown of OpenRouter on August 20, 2026, profiling the company as a unified inference layer that consolidates real-time data on how millions of developers route requests across 500+ models and 80+ providers as of August 2026. The report highlights OpenRouter's single-API approa The profile lists OpenRouter as founded in April 2023, headquartered in New York, with $153M total funding at Series B and around 100 employees, and cites COO Chris Clark on inference costs eventually eclipsing salaries as agentic adoption scales. It anchors the company inside a broader AI inference market projected to

OpenRouter's Fusion API routes prompts to multiple AI models simultaneously, claiming to outperform GPT-5.5 and Claude Opus 4.8 at half the cost of premium

OpenRouter has raised a $113 million Series B led by CapitalG. Its 5x growth in usage over six months indicates the multi-AI-model future is here.

OpenRouter, the AI model exchange, today announced a $113 million Series B led by Alphabet’s independent growth fund, CapitalG, with participation from inves...