GPT-5.6 SolvsGLM-4-9B
GPT-5.6 Sol | GLM-4-9B | |
|---|---|---|
| Specifications | ||
ParametersA rough measure of how big the model is. More parameters usually means more capable and more expensive to run, though it is a poor guide on its own — a smaller, newer model often beats a larger, older one. | — | 9B |
| API pricingUSD per 1M tokens · lower wins · base tier | ||
Input priceWhat you pay for everything you send the model — your question, plus any documents or earlier conversation you include with it. | $5.00 | — |
Output priceWhat you pay for the text the model writes back. It is normally the dearer half: producing an answer costs more than reading one. | $30.00 | — |
Cached input priceA reduced rate for text you send over and over. If every request starts with the same instructions or the same document, the provider keeps a copy ready and charges less to read it again. | $0.50 | — |
Cheapest inputLowest input rate across third-party providers, excluding the lab itself. The cheapest endpoint may run a quantised build or a shorter context — see "Available from" on the model page. | $5.00Azure | — |
Cheapest outputLowest output rate across third-party providers, excluding the lab itself. May come from a different provider than the cheapest input. | $30.00Azure | — |
| BenchmarksPublished by one model only | ||
BullshitBench v2Nonsense detection — Given a confidently-worded but nonsensical prompt, does the AI spot that it makes no sense and push back — instead of playing along and inventing an answer? The score is how often it clearly called out the nonsense. Higher is better. | 47% | — |
Gray Swan IPI · k = 1Prompt injection robustness — Attackers hide malicious instructions inside content the AI reads — a web page, an email, a document — and try to hijack what it does. Gray Swan's indirect prompt injection benchmark measures how often such an attack succeeds when the attacker gets a single try. Lower is better. | 3.1% | — |
Gray Swan IPI · k = 10Prompt injection robustness — Attackers hide malicious instructions inside content the AI reads — a web page, an email, a document — and try to hijack what it does. This variant gives the attacker 10 tries and counts an attack as successful if any of them works. Lower is better. | 16.3% | — |
Gray Swan IPI · k = 15Prompt injection robustness — Attackers hide malicious instructions inside content the AI reads — a web page, an email, a document — and try to hijack what it does. This variant gives the attacker 15 tries and counts an attack as successful if any of them works. Lower is better. | 20% | — |
CursorBench v3.2Agentic coding — Cursor's own test of harder, real-world coding tasks inside a code editor, on the refreshed v3.2 task set. Scores aren't comparable with v3.1. Higher is better. | 67.2% | — |
DeepSWE 1.1Agentic coding — Artificial Analysis' independent test of deep, agentic software-engineering work — the AI has to plan and carry out substantial coding tasks end to end. (Version 1.1 of the test.) Higher is better. | 73% | — |
Next.js EvalsNext.js coding — Vercel's open eval of how well AI coding agents build and migrate real Next.js apps — measured as the share of tasks the agent completes successfully. Higher is better. | 92% | — |
Supabase Evals · with skillsSupabase coding — Supabase's own open benchmark: a coding agent is dropped into a real Supabase project and asked to do real work — set up a schema, fix a broken security policy, debug an Edge Function — and every run is checked against a live Supabase stack. This is the headline number, where the agent has Supabase's own skills loaded, as most people building on Supabase would. The score is the share of scenarios it got right. Higher is better. | 95.5% | — |
Supabase Evals · no skillsSupabase coding — The same Supabase scenarios, but with none of Supabase's skills loaded — so it measures what the model already knows about building on Supabase, rather than how well it follows Supabase's supplied instructions. Higher is better. | 90.9% | — |
Frontier-Bench v0.1Agentic computer work — A hard, ever-evolving set of real computer tasks — coding, system administration, data work, and more — that an AI agent has to complete on its own. Run by the Harbor / Laude Institute team as the successor to Terminal-Bench (v0.1 is the first release of the task set). The score is the share of tasks solved. Higher is better. | 34.4% | — |
Terminal-Bench 2.1Agentic terminal coding — Can the AI work in a command-line terminal — running commands and finishing technical setup tasks the way a developer would? Higher is better. | 88.8% | — |
BU BenchBrowser agent — Can the AI drive a real web browser to finish tasks — clicking, filling forms, and navigating sites the way a person would? Run by Browser Use on their BU Bench task set. Higher is better. | 67% | — |
BrowseCompWeb browsing — Can the AI browse the web and track down hard-to-find answers? Higher is better. | 92.2% | — |
ARC-AGI-2Abstract reasoning — Puzzle-style tests of abstract reasoning and pattern-finding — the kind of thing people find easy but AIs often struggle with. Higher is better. | 92.5% | — |
GDPval-AA v2Knowledge work — economically valuable knowledge work (v2, re-based Elo) | 1748 | — |
Arena Elo (Text)Community preference — Real people chat with two anonymous AIs side by side and vote for the answer they prefer. Votes become a chess-style Elo rating on arena.ai — it measures which AI people actually like, not test scores. Higher is better. | 1481 | — |
Arena Elo (Code)Community preference (code) — Like the text arena, but people vote on which AI writes better code. The votes become a chess-style Elo rating on arena.ai. Higher is better. | 1620 | — |
Arena Elo (Vision)Community preference (vision) — Real people give two anonymous AIs the same image — a photo, a screenshot, a diagram — and vote for whichever reads it better. The votes become a chess-style Elo rating on arena.ai. It measures which AI people find more useful at looking at things, not how it scores on a fixed test set. Higher is better. | 1281 | — |
Arena Elo (Documents)Community preference (documents) — Real people hand two anonymous AIs the same PDF or document and vote for whichever answers better. The votes become a chess-style Elo rating on arena.ai. Unlike a fixed document benchmark, the files are whatever people actually brought along. Higher is better. | 1479 | — |
| Overview | ||
| Company | OpenAI | Z.ai |
| Release date | Jun 26 2026 | Jun 5 2024 |
| Access | Proprietary | Open Weight |
Other comparisons
GPT-5.6 SolvsClaude Opus 5GLM-4-9BvsClaude Opus 5GPT-5.6 SolvsGemini 3.7 FlashGLM-4-9BvsGemini 3.7 FlashGPT-5.6 SolvsMuse GlimmerGLM-4-9BvsMuse GlimmerGPT-5.6 SolvsGrok 4.6GLM-4-9BvsGrok 4.6GPT-5.6 SolvsDeepSeek-V4-Pro-0813GLM-4-9BvsDeepSeek-V4-Pro-0813GPT-5.6 SolvsMistral Medium 3.5GLM-4-9BvsMistral Medium 3.5
Frequently asked questions
GPT-5.6 Sol and GLM-4-9B don't publish scores on any of the same benchmarks, so there's no direct head-to-head comparison. Only GPT-5.6 Sol has a verified first-party API price: $5.00 per million input tokens and $30.00 per million output tokens. No pay-as-you-go API rate is tracked for GLM-4-9B. GLM-4-9B shipped 751 days before GPT-5.6 Sol, so benchmark comparisons should account for the intervening progress.
GPT-5.6 Sol is proprietary, while GLM-4-9B is open weight.
Direct benchmark comparisons are unavailable — GPT-5.6 Sol and GLM-4-9B don't publish scores on any of the same benchmarks.
GPT-5.6 Sol was released by OpenAI on Jun 26 2026.
GLM-4-9B was released by Z.ai on Jun 5 2024.
Only GPT-5.6 Sol has a verified first-party API price: $5.00 per million input tokens and $30.00 per million output tokens. No pay-as-you-go API rate is tracked for GLM-4-9B. Rates are pay-as-you-go API prices verified on August 18, 2026.
GPT-5.6 Sol is a proprietary model released by OpenAI. GLM-4-9B is an open weight model released by Z.ai.