In our ongoing testing, ReGrade cuts the cost per bug fixed by over 54%.

	Cost			Bug Fixes
Model↕	Without ReGradeBefore↕	With ReGradeAfter↕	Change▲	Without ReGradeBefore↕	With ReGradeAfter↕	Lab↕	Trials↕	Verdict
ling-2.6-1t	$0.09	$0.006	−93%	1.6	7.0	InclusionAI	10	✓
solar-pro4	$0.02	$0.002	−91%	3.8	10.6	Upstage	10	✓
qwen3.6-max-preview	$0.34	$0.05	−87%	5.7	13.8	Alibaba	8	✓
deepseek-v4-flash-0731	$0.03	$0.004	−86%	4.8	16.7	DeepSeek	22	✓
laguna-xs-2.1	$0.05	$0.007	−84%	1.8	10.3	Poolside	30	✓
deepseek-v4-pro-0813	$0.11	$0.02	−83%	6.8	13.6	DeepSeek	30	✓
minimax-m2.5	$0.02	$0.003	−83%	2.1	10.9	MiniMax	60	✓
kimi-k2.7-code	$0.05	$0.010	−81%	5.0	15.7	Moonshot	6	✓
minimax-m3	$0.06	$0.01	−80%	7.3	17.7	MiniMax	6	✓
sakana-namazu	$0.16	$0.04	−77%	5.0	12.0	Sakana	7	✓
seed-2.0-lite	$0.01	$0.003	−75%	2.9	14.7	ByteDance	60	✓
deepseek-v4-flash	$0.02	$0.005	−71%	3.8	7.0	DeepSeek	8	✓
haiku-4.5	$0.18	$0.05	−71%	2.5	12.8	Anthropic	60	✓
hy3	$0.02	$0.004	−71%	6.0	14.0	Tencent	9	✓
grok-4-1-fast-non-reasoning	$0.02	$0.005	−69%	2.5	9.7	xAI	21	✓
glm-5.2	$0.03	$0.008	−69%	6.7	16.6	Z.ai	39	✓
haiku-4.5 (Aug 2026)	$0.15	$0.05	−68%	2.7	14.5	Anthropic	20	✓
opus-4.6	$0.29	$0.10	−67%	7.0	17.7	Anthropic	6	✓
step-3.7-flash	$0.04	$0.01	−64%	3.4	11.0	StepFun	60	✓
qwen3.6-plus	$0.06	$0.02	−64%	4.0	16.0	Alibaba	7	✓
gpt-5.4-mini	$0.02	$0.009	−63%	3.6	13.2	OpenAI	10	✓
grok-4.20-0309-reasoning	$0.27	$0.10	−62%	3.2	12.3	xAI	21	✓
qwen3.8-max-0902	$0.21	$0.09	−58%	12.6	17.6	Alibaba	10	✓
fable-5	$0.59	$0.25	−57%	11.5	18.0	Anthropic	32	✓
gpt-5.6-luna	$0.01	$0.005	−57%	7.5	16.8	OpenAI	60	✓
glm-5.1	$0.05	$0.02	−56%	4.5	17.6	Z.ai	60	✓
fugu-ultra	$0.49	$0.21	−56%	10.2	18.0	Sakana	10	✓
qwen3-coder-plus	$0.07	$0.03	−55%	2.0	11.8	Alibaba	7	✓
muse-spark-1.2	$0.07	$0.03	−54%	12.0	17.0	Meta	6	✓
grok-4.5	$0.03	$0.01	−53%	11.2	18.0	xAI	8	✓
sonnet-5	$0.26	$0.13	−52%	5.5	17.4	Anthropic	60	✓
gpt-5.5	$0.17	$0.09	−50%	13.3	18.0	OpenAI	67	✓
muse-spark-1.1	$0.04	$0.02	−49%	13.2	17.7	Meta	7	✓
opus-4.8	$0.40	$0.21	−49%	9.1	17.9	Anthropic	60	✓
qwen3.7-max	$0.05	$0.03	−48%	7.3	16.7	Alibaba	6	✓
kat-coder-pro-v2.5	$0.05	$0.03	−46%	—	—	Kuaishou	5	✓
ling-3.0-flash	$0.004	$0.002	−45%	2.9	11.3	InclusionAI	29	✓
gemini-3-flash-preview	$0.02	$0.01	−45%	9.6	17.6	Google	47	✓
gpt-5.6-terra	$0.02	$0.01	−43%	12.0	17.9	OpenAI	60	✓
fable51	$0.25	$0.15	−38%	12.7	18.0	Anthropic	6	✓
kimi-k3	$0.06	$0.04	−37%	12.3	17.0	Moonshot	6	✓
grok-4.3	$0.02	$0.01	−35%	2.0	6.3	xAI	6	✓
opus-5	$0.25	$0.17	−31%	11.9	18.0	Anthropic	60	✓
longcat-2.0	$0.01	$0.008	−30%	6.7	11.0	Meituan	6	✓
gpt-5.6-sol	$0.07	$0.05	−29%	13.9	18.0	OpenAI	60	✓
gpt-5.4	$0.03	$0.02	−28%	13.9	17.9	OpenAI	79	✓
inkling	$0.03	$0.02	−21%	9.3	15.3	ThinkingMachines	6	✓
sonnet-4.6	$0.11	$0.09	−20%	7.3	17.6	Anthropic	87	✓
nemotron-3-ultra	$0.07	$0.06	−20%	3.0	13.8	Nvidia	10	✓
opus-4.7	$0.20	$0.17	−18%	10.6	17.8	Anthropic	84	✓
gemini-3.1-pro-preview	$0.12	$0.12	+2%	8.2	14.5	Google	8	*
mistral-medium-3.5	$0.05	$0.06	+20%	6.4	14.9	Mistral	20	*
qwen3-coder	$0.10	$0.12	+21%	1.7	5.0	Alibaba	15	*
gemini-2.5-flash	$0.008	$0.010	+30%	5.3	5.6	Google	15	*
grok-3-fast	$0.05	$0.07	+40%	6.3	5.5	xAI	21	*
deepseek-v4-pro	$0.01	$0.02	+44%	5.3	16.6	DeepSeek	8	*
grok-3-mini	$0.005	$0.009	+88%	1.9	1.9	xAI	16	*

✓ improved · * got worse · N/A: the model did not edit the file in the control run, so there is no baseline to compare against. Click a column to sort. Measured over 68 models, 22 labs, 1,571 agent-mode trials. Bug fixes are unaided (mean bugs fixed across the model's control-arm trials) and aided (the same average over the treatment trials), so the table can be sorted by how capable a model is on its own and by how far it gets with help. Both are our own measurement from the same trials as the rest of this table, not a third party's score. — means the model sits below the trial floor, so no figure is published for it.