diff --git a/providers/aki-io/models/deepseek-v4-flash-0731-284b.toml b/providers/aki-io/models/deepseek-v4-flash-0731-284b.toml new file mode 100644 index 00000000000..e5404865df3 --- /dev/null +++ b/providers/aki-io/models/deepseek-v4-flash-0731-284b.toml @@ -0,0 +1,13 @@ +# attachment are not supported on AKI.IO for security reasons and therefore are disabled +base_model = "deepseek/deepseek-v4-flash-0731" +reasoning_options = [{ type = "effort", values = ["none", "low", "high", "max"] }] + +# cost as described on the AKI.IO pricing page at https://aki.io/#pricing +[cost] +input = 0.20 +output = 0.50 +cache_read = 0.10 + +[limit] +context = 1_048_576 +output = 81_920 diff --git a/providers/aki-io/models/gemma4-26b.toml b/providers/aki-io/models/gemma4-26b.toml index 6b384af3386..75f503273f3 100644 --- a/providers/aki-io/models/gemma4-26b.toml +++ b/providers/aki-io/models/gemma4-26b.toml @@ -2,7 +2,6 @@ # AKI.IO uses the reasoning_option 'none' to disable reasoning base_model = "google/gemma-4-26b-a4b-it" attachment = false -reasoning = true reasoning_options = [{ type = "effort", values = ["none", "high"] }] [cost] diff --git a/providers/aki-io/models/kimi-k2.7-code-1100b.toml b/providers/aki-io/models/kimi-k2.7-code-1100b.toml index 463ce1b5147..a814eda61af 100644 --- a/providers/aki-io/models/kimi-k2.7-code-1100b.toml +++ b/providers/aki-io/models/kimi-k2.7-code-1100b.toml @@ -4,9 +4,11 @@ base_model = "moonshotai/kimi-k2.7-code" attachment = false reasoning_options = [] +# cost as described on the AKI.IO pricing page at https://aki.io/#pricing [cost] input = 0.86 output = 3.00 +cache_read = 0.18 [limit] context = 262_144 diff --git a/providers/aki-io/models/minimax-m2.5-230b.toml b/providers/aki-io/models/minimax-m2.5-230b.toml deleted file mode 100644 index bdfe92876ba..00000000000 --- a/providers/aki-io/models/minimax-m2.5-230b.toml +++ /dev/null @@ -1,12 +0,0 @@ -# attachment are not supported on AKI.IO for security reasons and therefore are disabled -base_model = "minimax/MiniMax-M2.5" -attachment = false -reasoning_options = [] - -[cost] -input = 0.25 -output = 1.20 - -[limit] -context = 196_608 -output = 65_536 diff --git a/providers/aki-io/models/qwen3.8-27b.toml b/providers/aki-io/models/qwen3.8-27b.toml new file mode 100644 index 00000000000..401df658ad4 --- /dev/null +++ b/providers/aki-io/models/qwen3.8-27b.toml @@ -0,0 +1,14 @@ +# attachment are not supported on AKI.IO for security reasons and therefore are disabled +# AKI.IO uses the reasoning_option 'none' to disable reasoning +base_model = "alibaba/qwen3.8-27b" +attachment = false +reasoning_options = [{ type = "effort", values = ["none", "low", "medium", "xhigh"] }] + +# cost as described on the AKI.IO pricing page at https://aki.io/#pricing +[cost] +input = 0.30 +output = 2.20 +cache_read = 0.10 + +[modalities] +input = ["text", "image"]