On July 8, xAI launched Grok 4.5. This is the first publicly released model following the merger of SpaceX and xAI, targeting developers, engineers, and enterprise knowledge workers with a focus on lower operational costs and faster response times.
Priced lower than similar products
Grok 4.5 is priced at $2 per million input tokens and $6 per million output tokens. For comparison, Anthropic’s Claude Opus 4.8 is priced at $5 per million input tokens and $25 per million output tokens; OpenAI’s GPT 5.6 Sol is priced at $5 per million input tokens and $30 per million output tokens.
Musk stated on X that Grok 4.5 is “roughly equivalent to Opus 4.7 but much faster,” indicating that xAI is positioning this model not to achieve maximum performance, but to emphasize a balance between capability, speed, and cost.
The programming test did not rank first.
Based on the benchmark results released by xAI, Grok 4.5 is competitive in several programming tests but does not lead. In the DeepSWE 1.1 test, Grok 4.5 scored 53%, below Claude Opus 4.8’s 59%, GPT 5.5’s 67%, and also trailing behind Claude Fable 5’s 70%.
On the SWE Bench Pro test, Grok 4.5 scored 64.7%, higher than GPT 5.5’s 58.6%, but still below Opus 4.8’s 69.2% and Fable 5’s 80.4%. The report noted that xAI’s comparison was still against GPT 5.5, not GPT 5.6, which was released later the same day.
Efficiency becomes the key selling point.
xAI places greater emphasis on model efficiency. For example, on the SWE Bench Pro benchmark, Grok 4.5 uses an average of 15,954 output tokens per task, while Opus 4.8 uses approximately 67,020 tokens—a difference of about 4.2 times. For teams that frequently invoke the model, this could lead to significantly higher overall costs.
The report also noted that Grok 4.5 generates output at approximately 80 tokens per second. xAI stated that the model’s training incorporated developer session data from the Cursor platform, including debugging traces and actual code modifications, rather than just static code repositories.
Grok 4.5 is currently available via API, Hermes, and the Grok build, supporting a 500,000-token context window. EU users cannot access it yet; xAI expects local access to be available by mid-July.