Nex AGI: Nex-N2.6

Nex-N2.6 became callable on Nex AGI’s own API on November 10, 2026, with the successor changing from Nex-N2.5 Pro’s Qwen3.5 mixture-of-experts checkpoint to a post-trained GLM-5.3-Flash base. The model has roughly 321B parameters, but the launch included no vendor post, readable model card, or benchmark table. The release was priced at $0.12 per million input tokens and $0.40 per million output, 60% above Nex-N2.5 Pro. It supports 262,144-token context, mandatory high or medium reasoning, text and image input, and function calling, while remaining served at fp8 precision by Nex AGI itself.

OpenRouter’s Nex AGI catalog lists Nex-N2.6 as a roughly 321B-parameter agentic model built on GLM-5.3-Flash, with a 262K-token context window. The listing says the model targets computer use, coding, and long-horizon tasks that involve tools, placing it as a tool-oriented successor within Nex AGI’s model lineup. The November 10, 2026 listing says Nex-N2.6 reduces repetitive reasoning and follows tools available in the current environment more closely than Nex-N2.5-Pro. The information is attributed to Nex AGI but appears on a third-party catalog without cited benchmarks or primary model documentation.

Recruitly’s recruitment benchmark for Nex AGI’s Nex n2.6 reports an aggregate score of 79 out of 100, with all 18 tested jobs passed. The evaluation covers JSON extraction, cover-letter generation, multi-turn tool calls, candidate ranking, long-document parsing, grounded replies, retrieval refusal, intent classification, and function calling. The harness reports a median completion time of 14.5 seconds and a measured cost per task of $0.0004. It identifies the model as open-weight, text-only, built on GLM-5.3-Flash with about 321B parameters, and operating with a 262,144-token context window.

Nex AGI’s Nex-N2.6 is listed as an agentic model built on GLM-5.3-Flash with about 321B parameters. The exact variant supports computer use, coding, and long-horizon tool tasks, and reportedly reduces repetitive reasoning while following available tools more closely than Nex-N2.5-Pro. Released November 10, 2026, the model offers a 262K context window and discounted API rates of $0.12 per million input tokens and $0.40 per million output tokens, with cache reads at $0.024 per million. The single listed host reports 90 tokens per second, 0.80-second P50 latency, and 100% uptime.

Nex AGI released Nex-N2.6 on November 10, 2026, positioning it as an agentic model for computer use, coding, and long-horizon tool tasks. The listed build is based on GLM-5.3-Flash with about 321B parameters, and the model supports text and image inputs with text output through an API. The page reports a 256K-token context window and pricing of $0.12 per million input tokens and $0.40 per million output tokens. It also characterizes the release as a major update, with pricing described as verified on November 10, 2026.