Gemini 4 Argon is now on LLM Gateway: here’s how it scores on voice tasks

Gemini 4 Argon became available through AssemblyAI’s LLM Gateway on October 13, 2026, offering Google Vertex routing, a 1,048,576-token context window, tool and structured-output support, and pricing of $2 per million input tokens and $10 per million output tokens globally. In the Voice Standard Bench, it achieved the highest dictation-cleanup result among more than 50 tested models, scoring 99.5, while its 87.2 overall score modestly exceeded Gemini 3.8 Flash’s 85.8 and trailed GPT-6 Luna’s 89.8. Argon was notably faster than Gemini 3.8 Flash across transcript lengths and particularly suited to high-accuracy dictation, but it cost substantially more than GPT-6 Luna and other near-comparable models, especially for long audio. GPT-6 Luna performed better for extraction, summarization, and correction on 15-minute and hour-long transcripts at roughly one-sixteenth of Argon’s average cost, though Argon returned results faster for longer inputs. The release recommends testing models against individual transcripts and prompts, with Argon accessible through an OpenAI-compatible endpoint using the model string gemini-4-argon.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.