Review · AISource: Moonshot AI
Kimi K3.1 Pro: hosted coding model with lower published rates than K3
Moonshot AI’s Kimi K3.1 Pro has lower published input and output rates than Kimi K3. [1] Moonshot describes Pro as further post-training of K3. It claims a one-million-token context window and improvements in coding and agent tasks. [2] Vercel lists Pro on AI Gateway with a release date of October 23, 2026. It documents AI SDK, Chat Completions, Responses and Anthropic Messages access. [3] Vercel lists vision, tool use and structured outputs for Pro. [3] Hosted access is available; Moonshot promises downloadable weights “by Nov 4.” [2]
AI coding model
Kimi K3.1 Pro
VerdictPromising
Proposed trial: Compare full-repository context with targeted retrieval in repository refactoring tasks using Kimi K3.1 Pro, measuring accepted patches and total billed cost.
- Best for
- Developers building coding agents who can compare context strategies using acceptance tests and spending per accepted task
- Pricing
- Published USD rates per million tokens: Pro costs $2.00 input, $10.00 output, $2.00 cache write and $0.20 cache hit; Highspeed costs $4.00 input, $20.00 output, $4.00 cache write and $0.20 cache hit. [1]
Pros
- One-million-token context window claimed by Moonshot [2]
- Lower published input and output rates than Kimi K3 [1]
- Tool use and structured outputs listed by Vercel [3]
- Multiple API interfaces through AI Gateway [3]
Cons
- Additional input costs on repeated uncached turns [1]
- Cache lifetime and accounting rules unspecified on the official homepage [1]
- Double base input and output rates for Highspeed [1]
- Downloadable weights still a future commitment [2]
Why it matters
Long context could hold code, configuration and tests together. Whether full-repository ingestion beats targeted retrieval on accepted patches or total billed cost remains unproven. At Pro’s published API pricing, 900,000 uncached input tokens cost $1.800 before output, cache writes, tools, retries and taxes. [1] Kimi K3 lists higher rates of $3 input and $15 output per million tokens. [1] Highspeed doubles Pro’s base input and output rates for what Moonshot claims is two to three times the speed. [1][2]
Attribution
Sources
- 1Moonshot AIKimi API Platform
- 2Moonshot AIKimi K3.1 Pro Tech Blog: Advancing Open Intelligence
- 3VercelKimi K3.1 Pro API, Pricing & Playground
- 4VercelKimi K3.1 Pro and Kimi K3.1 Pro Highspeed are available now in AI Gateway
Read the original source for full details and context.
AI
OpenAI adds optional 12-hour prompt caching to the Responses API
OpenAI added 12-hour prompt caching for requests that reuse context hours apart. Cache writes cost 2.5 times the ordinary input-token rate, making repeated reuse key to savings.
Review · AI
Exa 3.0 adds lower-cost Lite search and generally available Dynamic Highlights
Exa 3.0 adds Lite, a search tier priced below Auto, to Exa’s web search and content retrieval API for AI applications. Exa announced the update on October 19, 2026, and said it was immediately available through its API. The company says Dynamic Highlights is now generally available. It selects text across results to reduce duplication and allocate more space to useful sources, according to Exa. Exa also says it added a larger embedding model, expanded coverage and built Lite to use Auto’s index with less retrieval compute.
Article · AI
vLLM 0.32.0 adds structured decisions: how to compare them with constrained JSON
vLLM 0.32.0 enables an endpoint that reads model scores for predefined answers. A proposed document-triage test compares that approach with constrained JSON across accuracy, calibration, latency, and fallback handling.