News
ReleasesOVHcloud unveils Olala mini, its first in-house LLMChinaA 9B model beats a 1T model on FinQA, trained with Crusoe Managed RLReleasesThe internal metaphysics of an LLMReleasesMapping Language Model Internal Representations to Physical Laws and Scientific ReasoningResearchCan China keep rogue AI in check? Meet the start-up testing the top modelsFrontierMicrosoft aims to challenge OpenAI, Anthropic with cheaper, specialized AI modelsOpen modelsMicrosoft rolls out new first-party models and updates to agentic AI continuous improvementFrontierMicrosoft unveils AI models for coding, cybersecurity as it takes on Anthropic, OpenAIFrontierGoogle is making it easier to opt out of Gemini AI model training – here’s howFrontierOpenAI Unveils New Astra AI Model After Scrapping Prior ReleaseReleasesOVHcloud unveils Olala mini, its first in-house LLMChinaA 9B model beats a 1T model on FinQA, trained with Crusoe Managed RLReleasesThe internal metaphysics of an LLMReleasesMapping Language Model Internal Representations to Physical Laws and Scientific ReasoningResearchCan China keep rogue AI in check? Meet the start-up testing the top modelsFrontierMicrosoft aims to challenge OpenAI, Anthropic with cheaper, specialized AI modelsOpen modelsMicrosoft rolls out new first-party models and updates to agentic AI continuous improvementFrontierMicrosoft unveils AI models for coding, cybersecurity as it takes on Anthropic, OpenAIFrontierGoogle is making it easier to opt out of Gemini AI model training – here’s howFrontierOpenAI Unveils New Astra AI Model After Scrapping Prior Release
EveryLLM
Grok-1.5

Grok-1.5

xAI's long-context Grok model with stronger reasoning and coding skills

Grok-1.5 is a large language model from xAI, announced in 2024 as the successor to Grok-1. It extends the context window to 128,000 tokens and the company reports stronger results on maths and code benchmarks, with rollout starting for early testers and existing Grok users on X.

Share on WhatsApp

  • Context window of 128,000 tokens, up to 16 times the earlier length
  • 50.6% on MATH and 90% on GSM8K in xAI's tests
  • 74.1% on the HumanEval code generation benchmark
  • Perfect retrieval in Needle In A Haystack up to 128K tokens
  • Trained on a custom JAX, Rust and Kubernetes stack

What it is

Grok-1.5 is a large language model built by xAI, the company founded to work on AI. It was announced in 2024 as the next version of Grok-1, whose weights and network design had been released two weeks earlier. The company positions it as a general assistant model with better reasoning and problem-solving, and with a much longer memory for text. Its context window, meaning the amount of text it can hold in mind at once, reaches 128,000 tokens. xAI says the model handles long documents and complex prompts while still following instructions.

How it came about

xAI released the weights and architecture of Grok-1 as a snapshot of its work up to November of the previous year. Grok-1.5 followed as the improved model built since then. It was offered first to early testers and to people already using Grok on the X platform, with a wider rollout planned over the following days and new features promised alongside it.

Inside the model

xAI describes Grok-1.5 as trained on a custom distributed framework built from JAX, Rust and Kubernetes, running across large GPU clusters. A training orchestrator written by the team spots faulty nodes and removes them from the job automatically. The team also tuned checkpointing, data loading and job restarts to cut downtime when something fails. The company gives no parameter count, training data or chip details for the model itself.

What it can do

In xAI's own tests, Grok-1.5 reached 50.6% on the MATH benchmark and 90% on GSM8K, which cover school-level and competition maths, and 74.1% on HumanEval for code generation. The company reports maj@1 figures for the maths sets and pass@1 for HumanEval, and notes its GPT-4 comparison scores come from the March 2023 release. On the Needle In A Haystack retrieval test it found planted text perfectly across contexts up to 128K tokens.

How to use it

Grok-1.5 was to reach early testers and existing Grok users on the X platform first, then roll out more widely. xAI announced no downloadable weights for this version, unlike Grok-1.

Questions people ask about Grok-1.5

What is Grok-1.5?

It is a large language model from xAI, the follow-up to Grok-1. The company describes it as a general assistant model with improved reasoning, maths and coding ability, and a context window of 128,000 tokens for reading long documents.

How is it different from Grok-1?

xAI says reasoning and problem-solving improved, and the context window grew to 128K tokens, up to 16 times the previous length. Grok-1's weights and architecture had been published two weeks before the Grok-1.5 announcement.

How do I use it?

It was released first to early testers and people already using Grok on the X platform, with a wider rollout planned. xAI announced no public weights for this version.

How large is the model?

No parameter count is given. xAI describes only the training setup: a custom distributed framework using JAX, Rust and Kubernetes on large GPU clusters, with automatic ejection of faulty nodes.

Good reading and sources

    Where this page comes from

    There is no English Wikipedia article on Grok-1.5 yet, so the facts on this page come from the sources listed above: x.ai/news/grok-1.5. Some are the subject's own pages, so its claims about itself are reported as its claims. EveryLLM wrote the text in its own words, with AI assistance, and compares it with the sources to make sure no passage is copied.

    It may contain mistakes, and figures change over time. Nothing here is investment or buying advice. If something is wrong, tell us. Last refreshed 2026-11-14.

    The weekly digest

    Every Monday: the model of the day, new pages on EveryLLM, and the conferences and anniversaries of the week ahead.