Crypto news

A serious intrigue is brewing in the world of artificial intelligence. Chinese company Z.ai has released its flagship model, GLM-5.2, and crypto and tech communities are already buzzing with discussions about whether a true "killer" of Anthropic's Claude has emerged. Users, including well-known crypto blogger 0xMarioNawfal, claim the new model outperforms Claude in several scenarios while costing tens of times less to use. Let's break down what GLM-5.2 is and whether such bold claims are justified.

What is GLM-5.2 and why is it interesting

The developers position GLM-5.2 as a model for long and complex work sessions. Its key feature is a stable context window of 1 million tokens, which is five times larger than its predecessor, GLM-5.1. This allows the model to retain vast amounts of code or text in its field of view without losing quality.

Key features of the model:

• 1 million token context: Does not degrade during ultra-long sessions, allowing an entire codebase to fit into a single reasoning cycle.
• Two levels of reasoning enhancement: High for a balance of performance and token consumption, and Max for maximum power but with higher costs.
• Open MIT license: No regional restrictions, allowing the model to be run on your own hardware (self-hosting).
• API pricing: Remains at the level of the previous GLM-5.1 version, making it extremely attractive for developers.

Model parameters are available on HuggingFace and ModelScope. Local deployment is supported by frameworks such as transformers, vLLM, SGLang, and others. Access is also available through the GLM Coding Plan subscription and the ZCode desktop agent.

What the benchmarks show

According to Z.ai's own tests, GLM-5.2 is recognized as the strongest open model on the market. However, it generally falls short of the flagship Anthropic Claude Opus 4.8. On standard programming tests, the gap with GLM-5.1 is noticeable: 81.0 vs. 63.5 on Terminal-Bench 2.1 and 62.1 vs. 58.4 on SWE-bench Pro. Meanwhile, on Terminal-Bench 2.1, the score of 81.0 closely approaches Opus 4.8's 85.0 and surpasses Gemini 3.1 Pro's 74.0.

On long-horizon tasks, the picture is similar. On the FrontierSWE test, where the model manages open technical projects lasting tens of hours, GLM-5.2 lags behind Opus 4.8 by only 1%. However, it outperforms GPT-5.5 and the previous Opus 4.7 version. On PostTrainBench, which evaluates the improvement of other models through fine-tuning, GLM-5.2 surpasses Opus 4.7 and GPT-5.5, yielding only to Opus 4.8.

Thus, across all three tests, GLM-5.2 shows the best result among open models but still trails the top closed solutions from Anthropic and OpenAI.

How much does the AI cost and what's the catch

The GLM Coding Plan subscription is divided into three tiers with prices for annual payment at a 30% discount: Lite — $12.6/month, Pro — $50.4/month, Max — $112/month. Higher-tier plans provide priority access to flagship models and larger resource limits. Within the subscription, quota consumption depends on load: a 3x multiplier during peak hours (2:00 PM to 6:00 PM Beijing time) and 2x off-peak. Until the end of September, a promotion applies where off-peak usage is billed at 1x.

What users are saying

User reviews are divided. Strengths include recognition of the model as the strongest open neural network currently available, significantly improved basic logic compared to version 5.1, comparability with GPT-5.5 at a high reasoning level, and the ability to autonomously perform complex tasks. Users describe it as slow and expensive but extremely persistent in achieving its goal.

Criticism primarily targets the service and stability. The cloud infrastructure is described as extremely weak, with developers complaining about expensive pricing and poor support, noting that it's easier to pay for Claude or GPT. The neural network is also criticized for a tendency to get stuck in infinite loops and ignore commands. According to users, the model is tailored exclusively for benchmarks.

Separately, users note that the model only reveals its full potential in Max mode, which consumes several times more tokens than High mode.

So, is it a "Claude killer" or not?

There is no clear answer to this question. GLM-5.2 is recognized as the best open model for programming and autonomous tasks today. In certain long scenarios, it comes very close to Anthropic's flagship. The open MIT license, self-hosting capability, and low entry barrier make it a notable player.

However, it is bloggers, not benchmarks, who call the new model a "Claude killer." In most tests, Z.ai itself ranks its model below Opus 4.8. Additionally, users complain about unstable cloud infrastructure, high token consumption in Max mode, and poor support.

My analysis: GLM-5.2 is undoubtedly a step forward for open models and a serious signal for the market. It narrows the gap with the leaders but has not yet surpassed them. For developers who value openness and control, it is an excellent tool. But for those seeking a ready-made and stable solution, Claude or GPT remain a safer and more convenient choice. For now, GLM-5.2 is more of a "contender for the throne" than its rightful owner.