URGENT: GLM5.1 released and its Amazing (and cheap)

Z.ai dropped GLM 5.1 on a Friday night, and it scores 47.9 vs Opus 4.5's 45.3 on coding evals — for $7/mo on the yearly plan.

The Benchmarks That Matter

The headline number is the coding eval jump: GLM 5.1 hits 47.9, Opus 4.5 sits at 45.3, and GLM 5 from February only managed 35.4. Z.ai's own marketing copy literally brags about being "so close to Opus 4.5." Translation: they know they can't beat Anthropic on quality, so they're beating them on price.

Pricing Is the Actual Pitch

• Light plan: $10/month monthly, $7/month on yearly
• Comes with 3x the usage of Claude Code
• You can run it inside Claude Code right now — just change the model in your config to GLM 5.1

Boxmining frames it bluntly: $10 is "the cost of a McDonald's meal and you can make the app of your dreams."

One-Shot Coding in Practice

Boxmining built a Warhammer-themed Astro Invaders clone in one prompt. It hit a black screen, he said "Go fix it," and it self-corrected in the same session. Same setup also handled the SSL cert, Nginx config, and deployment without extra prompting. A Kanban board project ran cleanly with parallel agent teams.

What's Better, What's Worse

The big upgrade is agentic tool use and OpenClaw integration — the model now self-tests instead of just spitting code. The trade-off: a King AI review flagged that question-answering regressed compared to GLM 5. Z.ai clearly knew, which is why they marketed exclusively to coders. Expect slower speeds for the first few days as everyone hammers the new endpoint.

The China Factor

Zhipu (Z.ai's parent) was founded by Tsinghua University grads, already IPO'd, and trained GLM 5 on non-Nvidia Chinese chips. Weights for 5.1 are promised, so local self-hosting is on the table. GLM 5 launched in February 2025, making 5.1 a two-month turnaround. Competitors heating up: MiniMax 2.7, Qwen, Xiaomi MiMO V2 (free on Kilo Code), and ByteDance's image models.