2026-10-17
Vercel AI Gateway logs add a sessions view showing session affinity routing
AI Gateway logs now include a Sessions view that groups requests into sessions and shows how session affinity routed each one. Each session row lists the client app that ran it (such as Claude Code, Codex, or Cursor), along with cost, models, providers, and the share of input tokens served from cache across turns.
2026-10-16
Reflection AI's Beam now available on Vercel AI Gateway for free
Reflection AI's Beam, a US open-weight reasoning model for coding and agentic tasks previously in stealth, is now available on Vercel AI Gateway. Beam is free to use for a limited time during the launch period when served by Reflection AI.
2026-10-15
Vercel expands Enterprise Apps and Agents platform with governance, connectors, and AI Gateway features
Vercel published a blog post announcing expansions to its Enterprise Apps and Agents platform, covering a single deployment target for engineers and citizen developers, central governance via existing IdPs, scoped agent connectors (Vercel Connect), and AI Gateway access to hundreds of models with Zero Data Retention. The post cites customer examples Klaviyo and Mercor and reports usage figures from the vendor.
2026-10-15
Muse Spark Open now available on Vercel AI Gateway
Vercel announced that Muse Spark Open, an open-weight reasoning model from Meta, is now available on AI Gateway under the model name meta/muse-spark-open. The model supports text and image input with a 1M-token context window and is accessible via OpenAI-compatible and Anthropic-compatible APIs and coding agents such as Claude Code, Codex, and Cursor.
2026-10-13
Gemini 4 Argon now available on Vercel AI Gateway
Vercel announced that Google's Gemini 4 Argon model is available through Vercel AI Gateway under the model name google/gemini-4-argon. The model has a 1M token context window, accepts text, image, PDF, and video input, supports tool calling, and uses thinking levels (low, medium, high) instead of thinking budgets.
2026-09-28
Anthropic ships Claude Sonnet 5.5, near Opus 5.5 at half the price
Anthropic released Claude Sonnet 5.5 on September 28, 2026 at Sonnet 5's price of $2 per million input tokens and $10 output. Anthropic says it runs more than 30% faster than Sonnet 5 and uses up to 30% fewer tokens per task. It reports 70.6% on Terminal-Bench 4.0, against 10.3% for Sonnet 5. It is the first Sonnet with the cyber safeguards used on Opus 5.5, and higher-risk security tasks fall back to Sonnet 5.
2026-09-22
Anthropic launches Claude Opus 5.5 at 40% below Opus 5
Anthropic released Claude Opus 5.5 on September 22, 2026, the first model in its Claude 5.5 family. Anthropic says it performs at the level of Claude Fable 5.1 on most work and generates output more than 30% faster than Opus 5. Pricing drops to $4 per million input tokens and $20 output, from $5 and $25, and cache reads fall from $0.50 to $0.20. It reports 66.4% on Terminal-Bench 4.0 and 81.8% on OSWorld 2.1. Thinking can no longer be turned off, and biology and security tasks route to Opus 4.8 unless the account is in a verification program.
2026-09-01
Anthropic ships Claude Fable 5.1 and Mythos 5.1
Anthropic released Claude Fable 5.1 and Claude Mythos 5.1 on September 1, 2026. Fable 5.1 holds Fable 5's $10 per million input and $50 per million output, but cache reads drop to $0.25 per million, 75% less, which Anthropic estimates cuts typical workloads by about 25% and highly agentic ones by up to roughly 45%. Anthropic reports 55.8% on Terminal-Bench 4.0 for Fable 5.1 and 60.9% for Mythos 5.1, which is the same model with different safeguards, limited to vetted organizations.
