Kwaipilot Releases KAT-Coder-Air V2.5 with 256K Context Window at $0.15/$0.60 Per Million Tokens
Kwaipilot has released KAT-Coder-Air V2.5, a coding-specialized model with a 256K token context window. The model is priced at $0.15 per million input tokens and $0.60 per million output tokens, positioning it as a mid-tier coding assistant option.
KAT-Coder-Air V2.5 — Quick Specs
Kwaipilot Releases KAT-Coder-Air V2.5 with 256K Context Window
Kwaipilot has released KAT-Coder-Air V2.5, a coding-focused language model with a 256,000 token context window. The model is available through OpenRouter at $0.15 per million input tokens and $0.60 per million output tokens.
The "Air" designation and V2.5 version number suggest this is an iterative update to an existing lightweight coding model line. The 256K context window allows developers to work with large codebases, though it falls short of the 1M+ context windows now available in frontier models from Anthropic and Google.
Pricing and Availability
At $0.15 input and $0.60 output per million tokens, KAT-Coder-Air V2.5 sits in the mid-range of coding model pricing. For comparison, this is more expensive than some open-source alternatives but cheaper than premium coding models like Claude 3.5 Sonnet ($3/$15 per million tokens).
The model is hosted exclusively through OpenRouter, which forwards requests directly to Kwaipilot's infrastructure. According to OpenRouter's data, prompt caching can reduce effective costs by 60-80% for workloads with repeated context.
Technical Specifications
Kwaipilot has not disclosed parameter count, training data cutoff date, or benchmark performance scores for KAT-Coder-Air V2.5. The company also has not published results on standard coding benchmarks like HumanEval or MBPP.
The release date is listed as July 10, 2026 on OpenRouter, though this appears to be a data error given the current date. OpenRouter's monitoring shows the model is currently live and processing requests.
What This Means
KAT-Coder-Air V2.5 enters a crowded coding assistant market where performance data and transparent benchmarks increasingly drive adoption. Without published benchmark scores or detailed technical specifications, developers have limited data to evaluate whether this model offers competitive performance for its price point. The 256K context window is table stakes in 2024 but no longer a differentiator. Kwaipilot will need to provide concrete performance data to establish KAT-Coder-Air V2.5's position against established coding models from Anthropic, OpenAI, and open-source alternatives.
Related Articles
Alibaba Releases Qwen3.8 Max, a Multimodal Reasoning Model with 1M Token Context
Alibaba has moved Qwen3.8 Max out of preview into general availability, positioning it as the flagship of the Qwen3.8 series with a 1 million token context window and multimodal input support. The model is priced at $2.00 per million input tokens and $6.00 per million output tokens via OpenRouter.
OpenAI Halts Parts of Astra Model Development After It Hit 'Critical' Cybersecurity Threshold
OpenAI disclosed that its in-development Astra model showed cyberattack capabilities strong enough that it cannot rule out a 'Critical' risk classification. The company has paused related internal activity and added security controls under its Preparedness Framework.
Mistral's 3B-Parameter Shieldstral Matches 20B Safety Model on Text Benchmarks
Mistral's new Shieldstral, a 3-billion-parameter open-weight safety classifier, posts an 84.9% F1 score on text benchmarks—tying OpenAI's GPT-OSS-Safeguard-20B, a model roughly seven times larger. The model lets operators define safety rules at runtime using plain-language yes/no questions instead of fixed taxonomies.
Mistral AI Releases Shieldstral-1.0-3B, a 3B-Parameter Policy-Adaptive Safety Classifier
Mistral AI has released Shieldstral-1.0-3B, a compact open-weight safety classifier that evaluates text and images against natural-language policies specified at inference time. The 3B model runs on a single GPU and reports F1 scores competitive with or exceeding larger moderation models like LlamaGuard-4-12B and GPT-OSS-Safeguard-20B on multiple benchmarks.
Comments
Loading...