FactCheckRadar
Fact-check archive
Published fact-check
Claim checked
“MiniMax dropped M3 1M context 59% on SWE-Bench Pro close to opus 4.7”
Published
Updated
The claim that MiniMax M3 scored 59% on SWE-Bench Pro and came close to Opus 4.7 is supported by the company's official announcement and multiple tech outlets. However, the evidence shows this is a vendor-reported figure that has not yet been independently verified on public leaderboards. MiniMax's own blog states M3 "approaches Opus 4.7" on SWE-Bench Pro, and third-party coverage from Apidog and Lushbinary confirms the 59% score while noting independent confirmation is still pending.
MiniMax officially released M3 on June 1, 2026, and its blog post explicitly states the model scores 59.0% on SWE-Bench Pro, surpassing GPT-5.5 and Gemini 3.1 Pro while approaching Claude Opus 4.7. This matches the X post's claim exactly. Multiple tech publications including Lushbinary, Apidog, and Pasquale Pillitteri's site all report the same 59% figure, consistently framing it as vendor-reported data. Apidog's comparison piece specifically warns that "most of the numbers behind that claim come from MiniMax itself" and that "independent leaderboard confirmation is still pending." The claim about being "close to Opus 4.7" is supported by MiniMax's own framing, though the exact Opus 4.7 score isn't provided in these sources for direct comparison. The 1M context window claim is also confirmed across all sources.
The evidence consists of MiniMax's official blog post and several tech publication reports all dated June 1, 2026. While multiple sources corroborate the 59% SWE-Bench Pro figure, they all trace back to MiniMax's own announcement. No independent leaderboard verification exists yet. The sources are recent and directly relevant to the claim.
Medium
Your vote helps us see which fact-checks deserve more attention.
4 reviewed sources behind this verdict.
MiniMax's official announcement confirming the 59% SWE-Bench Pro score and 'approaches Opus 4.7' framing
Independent tech coverage confirming the score while noting it's vendor-reported and awaiting verification
Additional confirmation with context about market reaction and pending independent verification
MiniMax M3 launched June 1, 2026 as the first open-weights model to combine frontier coding, a 1M-token context window, and native multimodality. Full developer breakdown: the MSA sparse-attention architecture, 59% SWE-Bench Pro, 66% Terminal-Bench 2.1, 83.5 BrowseComp, $0.30/$1.20 promo pricing, how to access it, and where it fits in your stack. MiniMax shipped **M3** on June 1, 2026, and the pitch is unusually ambitious: the first open-weights model that combines frontier-level coding, a **1-m